Stripe

Bret Taylor of Sierra on AI agents, outcome-based pricing, and the OpenAI board

2713 summary words 12 min summary Watch video

Start with the signal

12 min read

Summary

At-a-Glance

  • Verdict: Watch fully
  • Core thesis: AI agents will create durable value when companies design context-rich harnesses around measurable business processes, rather than merely attaching models to existing software, APIs, departments, or per-seat pricing.
  • Why it matters: Taylor provides production evidence from Sierra, a concrete supervisor-model architecture, a critique of MCP-style orchestration, an outcome-pricing model, and an investment framework for how agents may shift value away from systems of engagement.
  • Best use: Use it to pressure-test Ken's agent architecture, workflow ownership, pricing, enterprise AI product strategy, and software investment theses.

Executive Summary

Taylor argues that coding agents work unusually well because repositories concentrate textual context, history, tests, compiler feedback, and documentation in one inspectable environment. General-purpose agents lack those advantages. His near-term prescription is to make other domains resemble code repositories: persist context in files, preserve intent and history, and build an agent harness containing documentation, skills, roles, tools, and operating instructions. He is increasingly skeptical that MCP servers and elaborate multi-agent diagrams alone provide enough shared context.

Sierra applies this approach to enterprise customer experience. Its agents operate across phone, chat, WhatsApp, websites, and apps, answer questions, and execute actions. Taylor says Sierra reached $100 million ARR in seven quarters, $150 million in eight, and roughly $165 million one month into the next quarter. Customers commonly begin with a narrow channel and use case, then expand; ambitious deployments such as Rocket Mortgage span search, sales, mortgage origination, and servicing, turning the agent into a branded digital front door rather than a support bot.

The production architecture is organized around customer journeys, goals, guardrails, tools, and information rather than rigid scripts. Sierra uses multiple models, including supervisors that inspect an acting model's reasoning and send work back when it violates policy or fails to retrieve authoritative information. Taylor presents this layered reasoning as more robust than prompt tricks, while acknowledging that individual technical advantages—such as Sierra's Cantonese voice support—will be commoditized and must eventually be replaced by product, workflow, distribution, and customer-domain advantages.

The broader strategic argument is that AI productivity should be managed at the level of end-to-end processes, not individual jobs or departments. Outcome-based pricing makes the vendor accountable for those processes, while agents that perform work may capture value historically held by systems of record and systems of engagement. Taylor expects durable ledgers to remain important, but sees greater risk for software whose moat is primarily a user interface, workflow layer, or consolidated data access. He remains bullish on applied AI because enterprises buy complete solutions from vendors aligned to specific buyers—not raw models—and believes enormous value remains available even if frontier-model development paused.

Key Takeaways

  • Claim: A context-rich agent harness may matter more than a polished protocol layer, and the best near-term general-agent architecture may resemble a source-code repository. | Evidence: Taylor contrasts OpenClaw's crude Markdown memory with blank-slate consumer assistants and notes that coding repositories provide textual context, version history, documentation, tests, compiler feedback, and familiar Unix tools such as grep. He suggests future SaaS products may expose an agent harness containing skills, documentation, roles, and expert operating instructions—not merely REST, GraphQL, or MCP endpoints. | Implication: Ken should treat persistent context, intent documentation, inspectability, and feedback loops as first-class control-plane components rather than assuming tool connectivity or sub-agent composition produces a capable agent. | Caveat: Taylor's skepticism concerns MCP as a sufficient architecture, not as a basic interoperability protocol; he also accepts browser automation, phone calls, and existing APIs as useful transitional rails.
  • Claim: Enterprise agents are moving from isolated support automation toward a company's primary branded interface across sales, service, and product usage. | Evidence: Sierra supports phone, chat, WhatsApp, web, and mobile channels; SiriusXM's agent Harmony serves both phone and website chat. Rocket Mortgage uses agents across Redfin home search, mortgage origination, and mortgage servicing. Taylor reports Sierra reached $100 million ARR in seven quarters, $150 million in eight, and approximately $165 million one month later. | Implication: Agent products should be designed around a durable customer relationship and shared cross-channel identity, memory, and capabilities—not as separate chat, voice, sales, and support installations. | Caveat: The median deployment still begins with one channel and a few bounded use cases, so the end-to-end digital-front-door model reflects the most ambitious customers rather than the typical starting state.
  • Claim: Reliable enterprise agents should be specified through goals and guardrails and monitored by layered supervisor models rather than encoded as rigid step sequences. | Evidence: Sierra's Agent Studio models end-to-end customer 'journeys' and identifies the tools and information required for each. If an acting model answers from memory when policy requires a lookup, a supervisor model inspects the reasoning, explains the violation, and sends the task back. Taylor illustrates the intuition by saying two 90%-effective layers can theoretically yield 99% effectiveness. | Implication: Ken should separate execution, policy supervision, evaluation, and recovery into distinct control functions while keeping business users focused on declarative goals and constraints. | Caveat: The 90%-to-99% illustration is explicitly simplistic and depends on errors not being correlated; the transcript does not provide audited production error rates or detailed security boundaries.
  • Claim: Outcome-based pricing is strategically different from usage-based pricing because tokens and compute are not reliable measures of customer value. | Evidence: Sierra charges a pre-negotiated amount when a customer-service case is resolved without human intervention and charges nothing when it escalates; sales agents can be priced through commissions. Taylor compares the shift to advertising moving from impressions to cost per click and argues that using one-hundredth as many tokens is irrelevant if an agent generates one-tenth as much GMV. | Implication: For agent businesses with measurable results, pricing should transfer implementation and efficiency accountability to the vendor and reward actual workflow improvement rather than seats, messages, or model consumption. | Caveat: Not every outcome is attributable or quickly measurable: a home-search interaction can be valuable even when no purchase occurs, so Sierra falls back to usage pricing where a defensible outcome metric is unavailable.
  • Claim: The atomic unit of AI productivity is an end-to-end business process, not a person, job title, or department. | Evidence: Taylor uses supplier onboarding as an example spanning legal, procurement, finance, IT, and a business owner. Instead of giving each department a copilot, a company could assign responsibility for reducing a hypothetical 17-day onboarding cycle to 17 hours by automating bounded steps and escalating only exceptions. | Implication: Ken should evaluate AI opportunities by mapping cross-functional processes, cycle times, decision rules, handoffs, and exception paths rather than by asking how many employees a generic assistant can replace. | Caveat: AI can absorb digital workflows much faster than physical operations; Taylor contrasts software and finance with flower shops, wet labs, clinical trials, and shipping, where physical execution or robotics remains necessary.
  • Claim: Agents may shift enterprise-software value from databases and systems of engagement toward optimized processes that perform valuable labor. | Evidence: Taylor distinguishes durable records such as ERP ledgers from engagement layers such as CRM interfaces and marketing workflows. An agent can reconcile information from three identity and CRM systems without first completing a major data-unification project, and an agent that generates leads, carves territories, audits financials, or reviews contracts may be more valuable than the database it reads. | Implication: Software diligence should separate true record integrity and regulatory moats from advantages based on interface ownership, data colocation, or manual workflow gravity. | Caveat: He frames this as an unresolved investment thesis, not a settled outcome; incumbents retain distribution, compliance, customer relationships, sales capacity, social proof, and the right to build strong agents themselves.
  • Claim: Applied AI companies can remain durable even as their current technical differentiators are absorbed into frontier models, but only if they convert temporary model work into enduring product and go-to-market advantages. | Evidence: Sierra built specialized Cantonese voice support for a bank's Hong Kong business even though Taylor expects nearly every voice model to support Cantonese within three years. He argues that enterprises buy complete solutions aligned to a CFO, chief customer officer, or chief digital officer, while model vendors sell to different buyers. He says trillions of dollars of economic value would remain unrealized even if model development paused. | Implication: Ken should value applied-agent companies on workflow depth, buyer alignment, deployment capability, proprietary evaluation data, distribution, and iteration speed—not on model access or a feature likely to become native. | Caveat: Taylor concedes that the software industry's fog of war is unusually thick and that Sierra itself could theoretically be compressed by future model capabilities.

Detailed Brief

Customer-experience economics produce expansion and second-order effects

  • Claims: AI support can automate a large majority of cases, but maximizing cost reduction is not necessarily the most valuable deployment objective.; Removing support friction can increase total interaction volume, improve customer listening, and turn service from a cost center into a retention or revenue mechanism.; The residual work sent to humans becomes more complex and potentially more fulfilling.
  • Evidence: Taylor says Ramp automates about 90% of cases and that well-executed deployments can reach roughly 70% to 90%.; One retailer's customer-conversation volume increased two to three times after replacing an unpleasant legacy chatbot; its total costs did not decline much, but the CEO valued the increase in customer contact.; SoFi reportedly gained 33 Net Promoter Score points following its Sierra deployment.; Taylor estimates a human phone interaction can cost $10 to $20, versus an AI interaction at roughly $0.10 to $0.20 and potentially $0.01 to $0.02 over time.; One client reported higher call-center employee satisfaction after simple cases were automated, even though average handle time rose because the remaining cases were harder.
  • Caveats: Automation rate alone can be misleading because case volume, case complexity, average handle time, satisfaction, retention, and revenue may move in opposite directions.; Taylor is Sierra's CEO, so the customer examples and company performance figures are management claims rather than independently validated results in the transcript.
  • Implications: Agent ROI models should include induced demand, churn reduction, customer intelligence, employee experience, and competitive response rather than only labor savings.; Once every competitor can automate the obvious workflow, the technology becomes an imperative; temporary advantage comes from using the lower marginal cost to redesign the customer experience faster.

AI-native organizational design favors empowered technical generalists

  • Claims: AI coding tools increase the leverage of tech leads and high-agency generalists who combine customer understanding, product taste, systems judgment, and the ability to debug.; The code itself becomes less precious as models produce more of it; durable artifacts increasingly include product intent, customer problems, requirements, tests, and documentation.; Organizations may become flatter because individuals can produce complete products with greater autonomy, although Taylor does not prescribe a settled structure.
  • Evidence: Taylor says he is trying to stop writing code by hand and stop treating source-code elegance as the primary object of craftsmanship, while continuing to care about correctness and robustness.; He proposes that an agent coding session should produce a product-management documentation artifact recording the intention, PRD, and customer problem in addition to code.; He cites Google and Facebook's emphasis on tech leads and product managers in product reviews rather than relying solely on engineering-management chains.; He describes a future product builder who combines taste, infrastructure knowledge, and deep customer understanding as potentially orders of magnitude more valuable than a narrow specialist.
  • Caveats: Taylor warns against optimizing for austere team size as a vanity metric: a one-person company can still lose to a ten-person competitor that invests more aggressively and captures the market.; The future role definition and reporting structure for these 'hyper-generalists' remain unclear.
  • Implications: Management layers justified mainly by information routing may weaken, while people who can own intent, judgment, and customer outcomes gain leverage.; Documentation quality may become a core production discipline because it supplies the durable context from which agents repeatedly regenerate implementations.

Governance observations and 2026 predictions

  • Claims: Taylor joined the OpenAI board after Sam Altman's removal as a mediator acceptable to both the existing board and Altman, rather than participating in the original decision.; He describes nonprofit governance as a fiduciary duty to a mission—ensuring AGI benefits humanity—rather than the conventional duty centered on shareholder value.; He predicts broader adoption of long-running agents, a visible AI-enabled scientific breakthrough, and a transition in which most Silicon Valley companies stop writing the majority of code by hand.
  • Evidence: Taylor says the post-crisis board had only three members and had to be rebuilt with representation spanning nonprofit purpose, safety, AI's economic effects, and infrastructure-financing expertise.; He compares the hoped-for scientific milestone with culturally legible moments such as Kasparov versus Deep Blue and AlphaGo, while noting that current mathematical results are too obscure for mainstream resonance.; He says the no-handwritten-code prediction would have sounded bold four months earlier but now feels plausible because coding-agent capability has changed so quickly.
  • Caveats: He explicitly expresses low confidence in the prediction of a mainstream scientific breakthrough.; His coding prediction is limited to Silicon Valley because tool diffusion across the broader economy will take longer.; The OpenAI discussion provides governance philosophy and board-composition context but little detail about specific contested decisions, safety procedures, or commercial structure.
  • Implications: Mission-governed AI entities require board composition that can evaluate safety, economics, and capital intensity simultaneously rather than treating governance as conventional technology-company oversight.; The speed with which coding practices changed over four months argues for short planning cycles and against locking operating assumptions into multi-year transformation plans.

Notable Concepts & Terms

  • Agent harness: The complete operating environment around an agent—context, documentation, skills, tools, roles, policies, tests, and instructions—rather than merely an API connection.
  • Harness engineering: The discipline of structuring context and feedback so an agent can perform reliable work; Taylor sees it as an emerging successor to narrow prompt engineering.
  • Journeys: Sierra's domain-specific representation of an end-to-end customer process, including the information, tools, goals, and constraints the agent needs.
  • Constellation of models: Sierra's use of multiple models with different responsibilities, including acting and supervising models, to improve reliability and policy adherence.
  • Goals and guardrails: A declarative specification that gives an agent freedom to reason toward an objective while constraining prohibited decisions and regulated behavior.
  • English over PSTN: Taylor's example of Sierra agents at healthcare organizations communicating with other agents in spoken English over the ordinary telephone network, demonstrating the value of existing universal rails.
  • Outcome-based pricing: Charging for a measurable business result such as a resolved case or completed sale, rather than for seats, tokens, messages, or raw utilization.
  • System of record of a process: Taylor's proposed category for agents that encode and execute an optimized business process, potentially capturing value previously concentrated in departmental databases.

Operator Notes / Why Ken Should Care

  • Require every proposed agent deployment to name one accountable process owner, a baseline cycle time or conversion metric, an exception policy, and a measurable target before implementation begins.
  • Run an architecture review comparing the current orchestration stack with a repository-style context layer that preserves intent, policies, prior decisions, tool documentation, and execution history.
  • Add adversarial evaluations for cases where a model believes it already knows a famous company's policy and therefore skips retrieval from the authoritative source.
  • Test whether supervisor failures correlate with executor failures before using multiplicative reliability assumptions in production forecasts.
  • Create a pricing matrix that separates attributable outcomes, proxy outcomes, and usage-only workflows; avoid forcing outcome pricing where attribution can be gamed or delayed.
  • In software investment reviews, score record integrity, regulatory necessity, distribution, and process ownership separately from interface usage and data colocation.
  • Treat temporary model gaps such as language coverage as acceleration opportunities, but set explicit sunset criteria so teams do not defend custom infrastructure after frontier models surpass it.
  • Exclude raw automation rate from executive scorecards unless paired with total demand, escalated-case complexity, customer satisfaction, retention, and revenue effects.

Source/Metadata

  • Title: Bret Taylor of Sierra on AI agents, outcome-based pricing, and the OpenAI board
  • Transcript words: 23603
  • Duration seconds: 6102
  • Timestamp note: No usable timestamps or chapter markers were present. The transcript includes substantial duplicated passages and a repeated appended interview segment.
Search this video Resume transcript at 0%
Full transcript 18138 words · 104 min read
0:01

SPEAKER_00

Brett Taylor is the ultimate Silicon Valley veteran. He was one of the creators of Google Maps, invented the Like button, was co-CEO of Salesforce. He pushed through Elon's acquisition of Twitter when he was on the Twitter board. He's now the chairman of the OpenAI board. And his day job is founder and CEO of Sierra, which is bringing AI to customer service. He's one of the smartest people I know on the topic of how AI is changing established companies. Cheers. Cheers.

0:25

SPEAKER_00

So most important question, have you installed OpenClaw on your work laptop? I have not. Have you played with OpenClaw? I have played with OpenClaw. I haven't bought a Mac Mini. You can put these things in virtual sandboxes for less money.

0:41

SPEAKER_01

It's really interesting. It's very compelling. It's probably the first. I wouldn't have predicted the first broad—I don't know if consumer is exactly accurate, but maybe a hobbyist use of AI would have been this kind of semi-rogue open source project that goes through three name changes in three days. Yes. And I love it. I love everything about the chaos of it, just because people in our circles have been talking about AI agents for consumer use and all these fancy computer using agents. And instead, you're chatting over WhatsApp with a thing on a Mac Mini that is mildly unhinged and insecure. It's fascinating. The whole thing is fascinating.

1:22

SPEAKER_01

But isn't that the thing that seems funny to me? If you look at the landscape, still in 2026, if you open a new Gemini chat or if you open a new chat, it's basically a blank slate, there's no memory. And then Claw people talk about the WhatsApp and Telegram integrations and things. But it feels to me a big part of the value is not only can it do stuff proactively, but it has memory. But the way it has memory is this super janky, it's movie Memento, it writes things to a markdown file. And it's just writing the things to remember. And the compaction is buggy. It doesn't always write down the exact right things to remember.

1:35

SPEAKER_01

But isn't it funny that you can get super polished mainstream consumer apps that have no memory at all, or this wildly insecure three name changes project that almost remembers things by scribbling notes in the margin? [SPEAKER_00] And that is the stage of consumer AI. I have a probably not very thoughtful but technical theory on this. [SPEAKER_00] Coding agents have gone through transformation over the past four months. The difference between October and now, our conversation about the future of software engineering would be materially different.

1:56

SPEAKER_01

[SPEAKER_00] And how often can you say that about a technology? People in my circles, you look at a coding agent and you extrapolate to other domains. [SPEAKER_00] You're wondering, could all digital tasks be like this? And the answer is obviously yes over some period of time. But it's really interesting because I think the hard part of engineering is in the details and code repos have very specific qualities.

2:19

SPEAKER_00

One is all the context is in one place in files that are largely textual, not binary. And for most broad information tasks, that's not true. When you're writing your Stripe annual letter, my guess is the sources of information are in so many different systems, data warehouses. And so it's not impossible for an agent to use those things. But the idea that you can straight line from coding agents to writing the Stripe annual letter, I don't totally buy.

2:27

SPEAKER_00

Yeah. And then similarly, when agents are actually performing work on a code base, there's feedback, there's compiler errors, there's often unit tests, there's integration tests, there's the history of every change made in a really formal format, along with code reviews. And so you can actually reflect. Maybe we as engineers have always modeled ourselves after robots. And now we can actually fully realize that vision. So what's interesting about it is the idea that I wrote a markdown file for memory. I think is maybe more significant than a hack. Actually, to some degree, turning your life into code.

3:05

SPEAKER_00

[SPEAKER_01] Yeah, it's everything in a file system that looks like source code, not because that's the only way these agents can work. But actually, it's a quite efficient way to get a mix of context and random access memory. [SPEAKER_01] If you think about a vector database, it's more random access, you have to know what to look for. But actually, that's not how real memory works. [SPEAKER_01] There's a mix of it. So you're loading a markdown file. And as you said, compaction, all these things matter. [SPEAKER_01] But the messiness of it actually probably produces a more useful agent than a lot of the fancier things.

4:05

SPEAKER_00

[SPEAKER_01] And I use memory and ChatGPT and I love it. [SPEAKER_01] But I actually think this idea that there's a directory of everything you've ever done is actually maybe more useful to an AI than people think.

4:16

SPEAKER_01

And actually, if you follow over the past couple months this emergence of harness engineering, where you're building the harness around an agent to do work. I wonder if in the short term, it might just be one of those idiosyncrasies of history, where mimicking a code base is actually the best way to make a general purpose agent work. Yes. And maybe over time we'll get fancier than that. But it's actually a relatively efficient harness for an agent. Maybe that's why. Yeah, and it's very terminal centric. And yeah, it's backward compatible. You can use grep. Yeah, exactly. You don't need to make some vector database.

5:02

SPEAKER_01

And the AIs really know how to use all the Unix tools. And so you get a lot of lift from that. That's exactly right. Software engineers were notorious for making tools for ourselves first. So then we just bend every other domain in digital towards that domain. But the reason I brought up things like unit tests, integration tests, OpenAI did a blog post on harness engineering. I can't remember which engineer did the post, but one of the more interesting parts was documentation. Rather than just having a single agent's markdown file, they had a directory of essentially the entire product, the architecture, and they're filling this out over time.

5:31

SPEAKER_01

[SPEAKER_00] And the agent's markdown became pointers to it. [SPEAKER_00] But my hypothesis, having used Codex a lot, I wonder if the output of a session where you make a change to Stripe's product should be a documentation artifact in addition to code, where the documentation artifact is actually the product manager version of John, and the code was the engineer version of John, where there's a lot in the code that is more transient. You might be fine to delete that. What was the intention? What was the PRD? What was the customer problem? Yes.

5:50

SPEAKER_01

It's actually the more durable asset. And I wrote this on X and one of the funniest comments would be it would be the greatest irony if software engineering agents made all of us just write documentation the whole time, just because notoriously every good engineer hates writing documentation. Now that's our job. But I don't know, it resonated with me. You might be fine to delete that. What was the intention? What was the PRD? What was the customer problem? Yes.

6:09

SPEAKER_01

It's actually the more durable asset. And I wrote this on X and one of the funniest comments would be it would be the greatest irony if software engineering agents made all of us just write documentation the whole time, just because notoriously every good engineer hates writing documentation. Now that's our job. But I don't know, it resonated with me.

6:17

SPEAKER_00

[SPEAKER_01] Where the documentation artifact is actually what the product manager version of John, and the code was the engineer version of John, where there's a lot in the code that is more transient. [SPEAKER_01] You might be fine to delete that. What was the intention? What was the PRD? What was the customer problem?

6:35

SPEAKER_01

Yes. [SPEAKER_00] [SPEAKER_01] It's actually the more durable asset. And I wrote this on X and one of the funniest comments would be it would be the greatest irony if software engineering agents made all of us just write documentation the whole time just because notoriously every good engineer hates writing documentation. Now that's our job. But I don't know, it resonated with me. [SPEAKER_00] [SPEAKER_01] How much are you AI code? You're a very prolific engineer in the old fashioned handspun way of writing code. And so how has that changed? The spoke artisanal code. Exactly. Yeah. Pour over. Yeah, pour over code. That's a really fine way to use that. [SPEAKER_00]

7:21

SPEAKER_01

I am trying to get to a world where I'm not writing code. It's hard emotionally, if that makes any sense. Mm-hm. I have a hard time not caring. I don't care about the assembly language produced by the compiler. So why should I care about the code? Why should I care about the code?

7:27

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] I care about correctness. I care about robustness. And I think, I know intellectually, I don't need to look at how the compiler unrolled this loop.

7:28

SPEAKER_01

[SPEAKER_00] Yeah. [SPEAKER_00] To verify its elegance and correctness. [SPEAKER_00] Yeah. Yet somehow I feel that way about code. And I'm not sure if the code doesn't matter, but I've been trying to force myself to not care because I feel like I won't be a self actualized software engineer in the future. Yeah. And maybe we'll just be like, of course Markdown is how we work with machines. Yeah.

7:49

SPEAKER_00

[SPEAKER_01] And so now that you're not writing the code, I really wonder what that programming system should feel like and look like. And I don't mind chatting with Codex, that's fine. [SPEAKER_01] But I also think, as you imagine all the tests that you care about, all the showing you demos and mockups. And I wonder what the future integrated development environment for lack of a better term will be in that world. [SPEAKER_01] So what I'm trying to do is force myself to not be emotionally attached to the code, which is very hard for me because that was my entire life. [SPEAKER_01] Yes.

8:02

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] Before I was proud of the elegance of the code that I wrote. But if I still care about the craftsmanship, what do I want? Right.

8:25

SPEAKER_01

And I haven't quite visualized it yet. It feels to me like a very interesting time in agentic engineering because you were talking about this domain of harness engineering and people having skills and MCP and everything. And it's always interesting when not only is the leading product in a category changing, we're just figuring out what the categories are that we need things for MCP or skills or stuff. And it's all very fast moving.

8:40

SPEAKER_01

[SPEAKER_00] And that just feels to me like a very interesting time where clearly a new way of engineering is shaking out. And 2026 is clearly not the final word. [SPEAKER_00] Absolutely. [SPEAKER_00] And in fact, I'm growing more skeptical of MCP as a meaningful part of the future. Not, it's fine as a protocol. [SPEAKER_00] Yeah. [SPEAKER_00] But essentially going back to your joke around OpenClaw just writing a big markdown file. I think it works better than a bunch of MCP servers.

8:58

SPEAKER_01

[SPEAKER_00] But going back to the point of every AI agent knows how to use all these things, I feel this view of a multi-agent world where you have all these agents that do tasks that are fraud detection. Yeah. And another one over here for personalization. And then you make a super agent, it does all these things. And it looks really good on a whiteboard, like the most elegant looking but completely nonsensical architectures. And then you realize if you anthropomorphize the Stripe experience and you're the checkout concierge. What information do you need to have as a priority to actually make that a humane experience?

9:14

SPEAKER_01

And what ends up happening in multi-agent systems is you stuff all the context in the sub agents and the one on top has no ability to actually not sound robotic. And then in contrast, you look at something like OpenClaw, it's just a bunch of markdown files and the memory kind of feels right, even though it's a little bit kludgy. And similarly, if you go back to my arguments about source control, a repo has so much context. So it's not that you just have the myopic view of the file you're editing, it really has some expansiveness.

9:22

SPEAKER_01

My sense is we're making true agents over time, the way we think about context and how that context is shared, so that the agent that's orchestrated actually understands what's behind all these APIs and why and the history will maybe look a little bit more like OpenClaw and less like MCP over time. And I think these agents need a lot more context than what MCP affords. Yes.

9:49

SPEAKER_00

[SPEAKER_01] One thing we've noticed is there's a bit of a what's old is new again phenomenon where with this agent of commerce stuff that's happening, we actually built the APIs for this ten years ago as part of that move of social shopping that was for a while. Buying on Instagram, buying on Twitter. [SPEAKER_01] Yeah. [SPEAKER_01] [SPEAKER_01] And there's just a feeling it didn't quite happen for a few different reasons at the time. But the concepts are very similar that you want some action at a distance, you want to be able to manipulate stuff off site.

10:04

SPEAKER_00

[SPEAKER_01] And similarly, I think Patrick has wanted for the longest time at Stripe the ability to just SSH into your Stripe account. You're like, what do you mean? [SPEAKER_01] It's a very ergonomic way for developers to work where you just be able to log into your Stripe account and you have a command line there. [SPEAKER_01] Yeah. [SPEAKER_01] And then you can list out all your payments or you can tail the payments log or you can. [SPEAKER_01] He wants tail and pipe and grab. [SPEAKER_01] Exactly. All these things.

10:30

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And of course, now we're building that because it's much more relevant in the agentic world. But I find, yeah, all the agentic stuff is also bringing back a lot of ideas that you might have had before. [SPEAKER_00] [SPEAKER_01] It's a very ergonomic way for developers to work where you just be able to log into your Stripe account and you have a command line there. [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] And then you can list out all your payments or you can tail the payments log or you can. [SPEAKER_00] [SPEAKER_01] He wants tail and pipe and grab. [SPEAKER_00] [SPEAKER_01] Exactly. All these things.

10:57

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And of course, now we're building that because it's much more relevant in the agentic world. But I find, yeah, all the agentic stuff is also bringing back a lot of ideas that you might have had before. Well, it is because to some degree, the elegance of Unix, which has been the basis of why everyone wants SSH and the curl command that was famously on the Stripe homepage. And then you can list out all your payments or you can tail the payments log or you can.

11:24

SPEAKER_00

[SPEAKER_01] He wants tail and pipe and grab.

11:30

SPEAKER_01

Exactly. All these things. And of course, now we're building that because it's much more relevant in the agentic world. But I find somehow, yeah, all the agentic stuff is also bringing back. I don't know if you have this experience as well. [SPEAKER_00] [SPEAKER_01] It's bringing back a lot of ideas that you might have had before. Well, it is because to some degree, the elegance of Unix, which has been the basis of why everyone wants SSH and the curl command that was famously on the Stripe homepage.

12:13

SPEAKER_01

[SPEAKER_00] For people who got it, it was remarkable because you could have all these tools that did something well that was small and useful and you could chain them together to make something great. [SPEAKER_00] I actually, I wonder in the future, we've talked a lot about this, if you look at the canonical software as a service application like Stripe's console, and obviously you have what the consumer sees, but the configuration that a Stripe customer will log into, you would have a web app. [SPEAKER_00] And that's the forms and fields and buttons and graphs. [SPEAKER_00] Yeah. And then you have the API.

12:41

SPEAKER_01

And it was typically a REST API or GraphQL API, and you could do stuff with it. And this is how computers talk to it, how humans used it. [SPEAKER_01] I wonder if the web application of the future will actually be, certainly you'll want a web app, for the rare human who wants to sign in, but will you have an agent harness? And what I mean by that is something more than the APIs, but just if you think about the harness that you provide in a code base, the skills, the documentation, the roles, imagine the person who's the greatest Stripe expert, who knows how to extract the most value from the Stripe account. That's the harness. Yes.

12:53

SPEAKER_01

Not the API, that's just the button you click. And will that be an endpoint on Stripe.com so that your agent knows how to get the most value from Stripe? Yes. And I imagine you don't care if your merchants log in. What you want them to do is drive value for themselves, drive GMV, drive payments.

13:12

SPEAKER_00

[SPEAKER_01] And so I'm really excited about that because an API is great. [SPEAKER_01] APIs are awesome, but a harness is basically the instruction manual for all the Unix commands that power Stripe. [SPEAKER_01] That's very interesting. [SPEAKER_01] Yes. [SPEAKER_01] And I think if you look at the shape of a lot of APIs that services have, and I don't think Stripe's API coverage is probably more complete than most, but ultimately it is a way to manipulate some of the highest value business objects in the thing.

13:42

SPEAKER_01

Whereas actually what you want is one, all of the data to be browsable in some identically accessible or textual way. [SPEAKER_00] [SPEAKER_01] And then all of the actions to be able to be taken by agents. [SPEAKER_00] [SPEAKER_01] And it turns out there's a lot of switches in the dashboard. [SPEAKER_00] Yeah. [SPEAKER_00] There's no API for it, and we are all as an industry collectively discovering that. [SPEAKER_00] And it might be easier to be a product manager in the future. [SPEAKER_00] Yeah. You just need to add the switch to the dashboard. [SPEAKER_00] You're like, yeah, it looks like a Russian submarine to switch this, but who cares, right?

14:35

SPEAKER_01

[SPEAKER_00] Agents can handle it. [SPEAKER_00] Yeah. [SPEAKER_00] And as long as the harness describes when to use that switch, it's easier than UI design in some ways. [SPEAKER_00] And that's fascinating to me. [SPEAKER_00] Yeah. [SPEAKER_00] But one funny point Dario made is it's not clear, well, there's a race between people getting their stuff accessible via agents and desktop computer use getting better. [SPEAKER_00] And so it's actually not clear will the approach be Stripe builds way more APIs, and that's how your agents manipulate the Stripe account, or you just give your agents access to Chrome and a login.

14:58

SPEAKER_01

[SPEAKER_00] Well, so actually I'll give a funny story here. [SPEAKER_00] So Sierra, my company. Sorry, we'll get to Sierra. [SPEAKER_00] No, it's fine.

15:30

SPEAKER_00

[SPEAKER_01] But there's a real funny story here. [SPEAKER_01] [SPEAKER_00] So Sierra powers a lot of healthcare companies. [SPEAKER_01] [SPEAKER_00] So on the healthcare payer side, health insurance. [SPEAKER_01] Well known for their API quality. [SPEAKER_01] [SPEAKER_00] Well, so first of all, they're actually pretty sophisticated engineers in these companies. [SPEAKER_01] [SPEAKER_00] I really enjoy working with them. So you end up with Cigna, Blue Cross Blue Shield on the healthcare payer side insurance. Then you have healthcare providers like Sutter Health that we work with. Then you have revenue cycle management.

15:45

SPEAKER_00

So R1 and revenue cycle management basically help providers get paid by the insurance companies. And then you have a lot of other people in the middle of pharmacies, PBMs, and they all call each other. So a healthcare provider has to call a payer because a procedure happened and they have to get paid. So we have payers with AI agents that pick up the phone. [SPEAKER_01] Oh, sure. And we have providers that have AI agents that pick up the phone and make phone calls. We have revenue cycle management companies that work to make outbound calls to do it. We've already had- [SPEAKER_01] Do they switch to the agent language? They don't.

16:28

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] Oh. [SPEAKER_00] We've done English over the publicly switched telephone network. [SPEAKER_00] [SPEAKER_01] So you have TCP IP and English over PSTN. [SPEAKER_00] [SPEAKER_01] And it goes, I mean, it sort of reinforces, I guess, Dario's point, which is you can engineer all these fancy protocols, but the rails that are already there already exist. [SPEAKER_00] And we have providers that have AI agents that pick up the phone and make phone calls. [SPEAKER_00] We have revenue cycle management companies that work to make outbound calls to do it. [SPEAKER_00] We've already had- Do they switch to the agent language? [SPEAKER_00] They don't. Oh.

16:53

SPEAKER_01

[SPEAKER_00] We've done English over the publicly switched telephone network. So you have TCP IP and English over PSTN. And it goes, it reinforces, I guess, Dario's point, which is you can engineer all these fancy protocols, but the rails that are already there already exist. [SPEAKER_00] Yes. And English is spoken by all AI agents. And the publicly switched telephone network has been around for 100 years and it all works. [SPEAKER_00] Yes. Which is fascinating. So you have all these fancy MCP things and we're doing English over PSTN.

17:26

SPEAKER_01

So on one hand, I think I actually agree with the principle that one of the powerful parts about AI with its ability to do text, do audio, and do what I'll call computer use, but you can call it a form of image recognition manipulation. Certainly that's useful because you get to the point where you don't need to fully finish the last mile to get value. The thing I'd say though, going back to your talk about all the actions and they're not all in the product, all the APIs don't exist.

17:30

SPEAKER_00

[SPEAKER_01] These visual interfaces were designed for us. [SPEAKER_01] So think of, Stripe is I think famously one of the few enterprise software companies with good design for a long time. [SPEAKER_01] And the Stripe dashboard is really elegant, right? [SPEAKER_01] And most enterprise software, you can't say that about their dashboards. [SPEAKER_01] I don't think the ideal agent harness will be that elegant because it's optimized for something else. [SPEAKER_01] Certainly that's useful because you get to the point where you don't need to fully finish the last mile to get value.

17:50

SPEAKER_00

[SPEAKER_01] The thing I'd say though, going back to your talk about all the actions and they're not all in the product, all the APIs don't exist.

17:50

SPEAKER_01

These visual interfaces were designed for us. So think of, Stripe is I think famously one of the few enterprise software companies with good design for a long time. And the Stripe dashboard is really elegant, right? And most enterprise software, you can't say that about their dashboards. [SPEAKER_01] I don't think the ideal agent harness will be that elegant because it's optimized for something else. [SPEAKER_00] It's optimized for the context that you need to perform complex multi-step procedures on behalf of a person. And my guess is it's just very different.

18:37

SPEAKER_01

And I think seeing the way you write a harness for a software or a coding agent, it's just so different than the way you do UI design. [SPEAKER_00] So I'm certain that it's great that you can click around a green screen, which is an oxymoron. But type around or click around a legacy on-premises enterprise software system.

18:50

SPEAKER_00

[SPEAKER_01] I think these harnesses will be really good. And I wonder if there is a world two years from now where Stripe's ability to work with the agent that manages commerce for a direct consumer e-commerce company, that would be one of the ways you're evaluated? [SPEAKER_00] Yeah. [SPEAKER_01] And in fact, if your harness is not compatible with the way their agents work, that's actually that you're not compatible with them.

19:18

SPEAKER_01

[SPEAKER_00] Yeah. And I'm not sure that's right.

19:26

SPEAKER_00

But I think it's great that these things are backwards compatible. [SPEAKER_01] It's great that our agents have spoken over the telephone already in English.

19:43

SPEAKER_01

That's really funny. But I don't think it's the long-term future because there's so much value that you can provide. And put another way, agents using a sophisticated application harness can just do a lot more and do a lot more with higher fidelity.

19:54

SPEAKER_00

[SPEAKER_01] Yes.

19:58

SPEAKER_01

Well, you see, we should get to Sierra because you see a lot of real world AI adoption. And so maybe start by grounding us. What is Sierra? The business has scaled very quickly. So one of the latest metrics that you can share because they keep changing from month to month as you grow? [SPEAKER_00] Yeah. So Sierra, we help companies make AI agents for their customer experience. So if you have a big phone line, these agents can replace your IVR system and pick up the phone. If you have a digital chat system, an AI agent can pick it up. You don't need to wait on hold.

20:29

SPEAKER_00

Yeah.

20:30

SPEAKER_01

These agents can not only answer questions but take action on your behalf. We work with healthcare companies like Cigna. We just did a great case study with SoFi. And I'm really proud that we raised their Net Promoter Score by 33 points just because it's so delightful. Wow. It's really fun to see all these different brands across a wide range of industries get so much value from their agent. We're the leader in the space. You talked about the metrics. We reached $100 million in ARR in seven quarters, $150 in eight quarters. We're around $165 now one month into our next quarter. So growing really rapidly and really proud of the momentum that we have. That's super cool.

21:03

SPEAKER_01

[SPEAKER_01] What is the typical adoption? Are people using it for email chat support because that's the easiest modality? [SPEAKER_01] Do they adopt it for everything, including phone and stuff? [SPEAKER_00] It's changed a lot over the past two years. But I'll say the median customer, and they'll describe some interesting outliers, which I hope are glimpses of the future. So most will start with one channel and a few use cases. So at a lot of healthcare companies, phone remains the dominant channel. So say, for a few types of phone calls, let's have the AI agent take them and see how it does. Do people like it? Are people comfortable with it? Does it lower our cost?

21:41

SPEAKER_01

Does it raise whatever metrics? Usually it's customer satisfaction. And does it work more effectively? So for example, for a car insurance company, it'll be first notice of loss. I got in a fender bender, and that would be the typical way you start. For a lot of more digitally native companies, they'll start with chat and something similar. [SPEAKER_00] But almost all of our clients will do both. So SiriusXM, if you call them on the phone, their AI agent Harmony, which I love that name for SiriusXM, will pick up the phone. And if you go to their homepage and you see the chat, that's also the same agent. Does it raise whatever metrics? Usually it's customer satisfaction.

22:20

SPEAKER_01

And does it work more effectively?

22:28

SPEAKER_00

[SPEAKER_01] So for example, for a car insurance company, it'll be like first notice of loss.

22:31

SPEAKER_01

I got in a fender bender, and that would be the typical way you start. For a lot of more digitally native companies, they'll start with chat and something similar. [SPEAKER_00] But almost all of our clients will do both.

22:49

SPEAKER_00

[SPEAKER_01] So SiriusXM, if you call them on the phone, their AI agent Harmony, which I love that name for SiriusXM, will pick up the phone.

22:51

SPEAKER_01

And if you go to their homepage and you see the chat, that's also the same agent. So the neat part is, I think it's pretty neat because you have all of your customer experience team or whatever you might call it at your company.

22:53

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] They can spend all their time on one thing. [SPEAKER_01] Yes. [SPEAKER_01] And it actually works over WhatsApp. [SPEAKER_01] [SPEAKER_00] It can work online. [SPEAKER_01] [SPEAKER_00] Yes. [SPEAKER_01] [SPEAKER_00] It can work on your website, work in mobile app.

23:05

SPEAKER_01

[SPEAKER_00] It can pick up the phone. [SPEAKER_00] That's a pretty big change. [SPEAKER_00] A lot of our clients, when we start working with them, they'll have a digital team and a call center team and all these different teams. And we've gotten to the point because we've digitized the last remaining analog channel, which is the telephone. Those are all unified. When I start with a glimpse of the future, we have a few really ambitious customers like Rocket Mortgage, a great Detroit company. They own Redfin. They bought a mortgage services company called Mr. Cooper. If you go to Redfin, you can search for a home using an AI agent.

23:39

[SPEAKER_01] If you go to rocket.com, you can originate a mortgage with an AI agent and you can service that mortgage. [SPEAKER_01] It becomes a product usage rather than just customer service. [SPEAKER_00] [SPEAKER_01] And really end to end, sales, service. [SPEAKER_00] And I think that's really exciting. I mean, our whole view is that if we're in 1994 and you were talking about this internet phenomenon. I wouldn't have fun. [SPEAKER_00] [SPEAKER_01] I was a bit young, but yeah.

23:56

SPEAKER_00

[SPEAKER_01] Yeah. [SPEAKER_01] I would have my Nirvana shirt on. [SPEAKER_01] Yeah, exactly. We would be talking about, look, this is your digital front door.

24:04

SPEAKER_01

[SPEAKER_00] Or maybe we wouldn't have the language to say that. [SPEAKER_00] [SPEAKER_01] It becomes a product usage rather than just customer service. [SPEAKER_00] [SPEAKER_01] And really end to end, sales, service. [SPEAKER_00] And I think that's really exciting. Our whole view is that if we're in 1994 and you were doing Cheeky Pint about this internet phenomenon. I wouldn't have fun. I was a bit young, but yeah. I would have my Nirvana shirt on. [SPEAKER_00] We would be talking about, look, this is your digital front door. Or maybe we wouldn't have the words to say that. On the information superhighway. [SPEAKER_00] On the information superhighway.

24:26

SPEAKER_01

And I think the same is true of most companies, AI agent singular. There are lots of agents, but the one with your brand at the top that your customers interact with is special. And that's the one we're trying to power for companies. So you think this, what customer is built on Sierra, your aspiration is that it just becomes sometimes the primary way people deal with the company. I think a company's AI agent will be the vast majority of their digital interactions. And I think digital has come to include the telephone. And that's a big shift because we think of that differently.

24:32

SPEAKER_01

Yes. And that's a huge change just because the bigger shift. So customer service, which is one big part of what we do, but not the only thing we do, is traditionally been thought of as a cost center because it's really expensive. Yeah. So I'm sure you have people answering the phone for your clients. And depending on where they're located and how well trained they have to be, how simple is the case, it can cost $10, $20.

24:38

SPEAKER_01

Yeah. It can be much less if it's a more simple case. And you know, you have some customers who pay you millions of dollars and you'll answer the phone anytime they call. And you might have one that has not even started monetizing yet. And you might want to call them, but there's a limit to how much you can afford to talk to that person and still have a profitable business. I always joke, it's probably easier for you and me to call Sundar than to get Google customer service on the phone.

24:39

SPEAKER_01

It's very hard to get Google customer service on the phone. But it's not because they don't like you. It's just if you think about the average revenue per user of Google, they literally can't afford to do it. [SPEAKER_00] [SPEAKER_01] Yes. So now if you take that $10 or $20 phone call and you make it $0.10 or $0.20 and over time $0.01 to $0.02, all of a sudden not only can you afford to provide a great customer experience to more people, even less profitable customers or in lower margin businesses, which I think is very exciting. So it's not just doing what you did before but new.

24:51

SPEAKER_01

You can provide better customer service. You really can. And then just think about running a subscription business where you care as much about customer acquisition, you care a lot about churn because that's how your lifetime value equation works. Yes. And you think about, okay, if I had a budget of how much I spent on service and now I can do 100 conversations more than I could before, can I actually reduce my churn rate? Yeah. [SPEAKER_00] Can I improve lifetime value? And then the interesting thing is then you realize that, wow, all of my competitors have access to the same technology. Yes. [SPEAKER_00] [SPEAKER_01] Yes.

25:16

SPEAKER_01

[SPEAKER_00] And they're saying, okay, what are my competitors going to do to actually take my customers away from me? And then that's where you start to get things. You know, the ATM machine didn't actually reduce bank branches because some bank had the great idea of I'm going to put different people in this branch. [SPEAKER_00] Right. [SPEAKER_00] [SPEAKER_01] It'll generate revenue. And all of a sudden it wasn't job displacement, but something completely different.

25:30

SPEAKER_01

So I think the exciting part in our world is you're taking something that was so expensive that people literally hid their phone numbers. Yeah. So people couldn't call them and you're making it inexpensive and delightful. And everyone, the thing I'm excited about is the second and third order effects are going to be really interesting and very hard to predict. And that's pretty exciting.

25:39

SPEAKER_01

I want to come back to this idea of the agent as the UI, because I found it really interesting. We talked about this a bit in our newsletter in the context of agentic commerce, where again, I think people are trying to pitch too much of the end state of, you know, fully autonomous, you know, the robots just choosing few. And the point we always make is, let's just start with not having to fill out the web form. No one likes filling out forms on the internet. Speak for yourself.

25:58

SPEAKER_01

I want to come back to this idea of the agent as the UI, because I found it really interesting. We talked about this a bit in our newsletter in the context of agentic commerce, where again, I think people are trying to pitch too much of the end state of fully autonomous robots just choosing. And the point we always make is, let's just start with not having to fill out the web form. No one likes filling out forms on the internet. Speak for yourself.

26:02

SPEAKER_01

[SPEAKER_00] I wonder if using websites will have been actually a bit of a fax machine. We used emails over telephone lines as a way of transmitting information or there was an era of voicemail memos for you ever in the working world for that? Some people still do this where they do the voice. But companies would blast a voicemail memo to employees at the company. And that's a way of distributing information. And all these things are very moment in time and maybe navigating websites and filling out forms was a bit of a moment in time. Is that how you see things playing out?

26:12

SPEAKER_01

I don't know. It's interesting because if you look at the past few iterations of technology, you had the PC revolution, then you had the internet and the browser, then the smartphone came out and the tablet. And I was more optimistic about tablets than the way the world turned out. You know, I see more tablets on airplanes, but I don't, I'm guessing if I walked around Stripe, I would see very few tablets.

26:21

SPEAKER_01

[SPEAKER_00] Yeah. And similarly, there's more smartphones than people, but there's still about 2 billion PCs in the world. And I think it peaked some number of years ago, but it hasn't gone down as far as I know. That's interesting, right? We added to our digital world, but I think the more interesting metric is for you and me, what percentage of emails were sent through each device.

26:22

SPEAKER_00

[SPEAKER_01] And certainly from 2010 to 2020, most of the world might have transitioned from percentage of email on desktop to smartphone significantly. And so it's almost like a market share of digital interactions, which I think is a really interesting way to think about it.

26:24

SPEAKER_01

[SPEAKER_00] And certainly as you think of Stripe's business, where does commerce originate? You saw that move to mobile, but it doesn't mean that people, it's actually a very big thing. If you just, you wouldn't want to eliminate the PC commerce business. That would actually be catastrophically bad. [SPEAKER_00] Yeah.

26:30

SPEAKER_00

[SPEAKER_01] And so then you look at AI agents and I believe most businesses will have them as their primary digital interface. And it's because it works over WhatsApp and it works over the phone. Yeah. [SPEAKER_01] If smart speakers make a comeback, they'll work over smart speakers. Which they may now.

26:46

SPEAKER_01

[SPEAKER_00] Yeah, totally. Maybe smart speakers were just too early.

26:47

SPEAKER_00

[SPEAKER_01] Yeah. Well, asking for the weather just turns out to be not the biggest market in the world.

26:51

SPEAKER_01

[SPEAKER_00] Exactly. But now— Set a timer.

27:00

SPEAKER_00

Set a timer.

27:04

SPEAKER_01

Yeah. I mean, it's amazing they've made that much money off a timer setting speaker. And so it is very future proofed. [SPEAKER_00] Yeah.

27:14

SPEAKER_01

Because it's fundamentally conversational. [SPEAKER_00] But it's going from punch cards, mice and keyboards, touch screens, now voice and chat and probably 3D immersive at some point. Does it just add and make the other ones less important is probably the way I think about it. I do wonder if we'll see the end of the smartphone at some point. [SPEAKER_00] It doesn't seem anywhere close right now. But it is interesting. I mean, I think most people don't love how much we're addicted to staring at this glowing screen. Yes.

27:36

SPEAKER_00

On the other hand, you can't talk to TikTok.

27:48

SPEAKER_01

Right. It's fundamentally visual. But I wonder if there's a world where you could actually be really productive without such an invasive device on your body.

27:56

SPEAKER_00

Yes.

27:59

SPEAKER_01

And if that's the case, can it offer an opportunity to unwedge some of the addictive properties of these technologies and get a lot of the benefits from it? [SPEAKER_00] Because for me, I think all of us are so connected, you end up checking your email. And then you're like, where have I been for the past hour? Yeah. [SPEAKER_00] And the fact that we actually have technology that affords that innovation now, I think that's quite interesting. So I don't know what the future is, but I'm very excited for it. I know it sounds really cheesy, but we've now changed the ingredients available and we have a lot more recipes we can cook. And I think that's very exciting.

28:13

SPEAKER_01

I agree that I'm excited for not having to look at the screen for as many things for a variety of reasons. When a customer installs Sierra, I know there's a significant customer satisfaction component as well as cost, but I'm curious what kind of cost difference does it make? And maybe relatedly, when a customer is fully deployed, what kind of mix do they see between queries fully resolved agentically, things that end up having a human who is presumably somewhat AI assisted, but what does a normal equilibrium look like there?

28:14

SPEAKER_01

[SPEAKER_00] Yeah. It turns out most of our clients have pretty different priorities. Some are very focused on cost savings and you can automate very high percentages of your cases. There's a company called Ramp that's a really impressive tech firm. We had Eric just here. Oh, that's great.

28:33

SPEAKER_01

[SPEAKER_00] Well, they're automating 90% of their cases. They're really sophisticated though, because they're getting in front of cases before they escalate. But I think it's an example of a really fantastic company implemented really well. And you can see anywhere between 70, 90%, which is really incredible. The interesting thing though is there's counterintuitive effects to it. The cases that do make it to your customer service team can end up more complex by definition. Yeah, yeah. [SPEAKER_00] What's called average handle time will actually go up. Oh, that's great. Well, they're automating 90% of their cases.

28:50

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] They're really sophisticated though, because they're getting in front of cases before they escalate. [SPEAKER_00] [SPEAKER_01] But I think it's an example of a really fantastic company, implemented really well.

29:01

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] And you can see anywhere between 70, 90%, which is really incredible. [SPEAKER_01] The interesting thing though is there's counterintuitive effects to it.

29:20

SPEAKER_01

The cases that do make it to your customer service team can end up more complex by definition. Yeah, yeah. [SPEAKER_00] What's called average handle time will actually go up. [SPEAKER_00] Yes.

29:36

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] And we heard from one of our clients that actually their satisfaction of their call center agents went way up too, because it turns out it's way more fulfilling.

29:41

SPEAKER_01

Totally. To solve a hard problem. Have you tried plugging it out and plugging it back in again? Exactly. The other interesting one, we had one retailer whose volume, total volume, went up almost as much as they saved from the AI agent. Because- Jevon's paradox. It was a form of that. So if you've used a chatbot from three years ago, they were really annoying. Three years ago, if you said, do you like chatbots? There was zero people who would say yes. It's so funny that there was a Silicon Valley wave of hype around chatbots. It was even earlier than that. It was 2018. [SPEAKER_00] [SPEAKER_01] It was pre-LLMs, pre-transformers. Yeah.

30:06

SPEAKER_00

[SPEAKER_01] And they were just multiple choice machines or something.

30:06

SPEAKER_01

It was just the worst products of all time. [SPEAKER_00] And so replacing it with something that was a delightful way, people are like, I'm going to talk to this thing a lot more. [SPEAKER_00] So they ended up keeping it, their cost didn't really go down, but their volume of customer conversations went up two or three X.

30:24

SPEAKER_00

Yes. And the CEO was incredibly happy about it. They were like, we're now actually listening to our clients. [SPEAKER_01] So it sounds funny, but it's a choice how much you want to drive cost savings with AI versus other metrics. [SPEAKER_01] Most of our clients are interested in the top line metrics. [SPEAKER_01] And so if you could save $10 or save $1 and improve your net promoter score and competitive positioning by a meaningful amount, everyone in the world would choose the latter. So they ended up keeping their cost didn't really go down, but their volume of customer conversations went up two or three X.

30:41

SPEAKER_01

[SPEAKER_00] Yes.

30:43

SPEAKER_00

And the CEO was incredibly happy about it. They were like, we're now actually listening to our clients.

30:51

SPEAKER_00

[SPEAKER_01] So it sounds funny, but it's a choice how much you want to drive cost savings with AI versus other metrics. Most of our clients are interested in the top line metrics. And so if you could save $10 or save $1 and improve your net promoter score and competitive positioning by a meaningful amount, everyone in the world would choose the latter. So that's the interesting thing going on right now because going back to about 1994 and we're hawking websites on this podcast. I think if we were to go to a major bank and say, if you launch a website, you're going to have a competitive advantage against every other bank.

30:59

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] With the benefit of hindsight, that would have been overpromising. The correct thing to say was, if you don't launch a website, you have to have a website.

31:05

SPEAKER_01

And so this technology is broadly available. Yes. [SPEAKER_00] [SPEAKER_01] And so as a consequence, you can't just launch in all parts of AI, not just Arbus. You can't just launch AI, absorb the cost savings, pass it on to shareholders unless you have a monopoly. [SPEAKER_00] Yes, yes. But most businesses, it's a consumer surplus.

31:08

SPEAKER_00

[SPEAKER_01] Exactly. So you're either going to lower prices, but I think that's why it's an overused analogy. But the ATM bank branch thing is really interesting because if every single company in an industry has access to technology, I would say it's an imperative, not a competitive advantage.

31:09

SPEAKER_01

Yes. And the more interesting board discussion is when everyone adopts the obvious things. Yeah. [SPEAKER_00] Customer experience, customer service, software engineering, legal. Just pick the ones where there's solutions off the shelf, solutions available now. Yeah. What will the industry look like? And my guess is you could ask how GPT think. And my guess is there's some really interesting second order effects. [SPEAKER_00] And when you have competitive markets, you're going to end up investing, lowering prices, whatever it may be. And that's the thing I don't think it's talked about enough. And I actually think that we just, it happens with every technology change.

31:46

SPEAKER_01

[SPEAKER_00] You project it through the lens of what you're doing today. [SPEAKER_00] Yes. And you don't take an effect. It's a multiplayer game that we're all in right now. Yes. And that's fascinating to me. [SPEAKER_00] And so the change is disruptive, but I think it's going to be like, I'm very excited for the next few years as the world absorbs the technology. We start getting to some of the second and maybe even third order effects. What is the most impressive AI adoption or AI native behavior you've seen from a client?

32:19

SPEAKER_01

[SPEAKER_00] Oh, that's a really good question. I'll probably say Rocket where we have a really great relationship. I think Varun is their CEO, Sean Mahochra is their CTO.

32:19

SPEAKER_01

Two people who are not only curious about AI, but very interested in transforming the home ownership experience with AI. And I don't know when you got your first mortgage, but it's super intimidating. It's not a modern process. They literally call it mortgage folders for a reason. Like it used to be a folder. And for me, it's an example of a company trying to transform an industry. And the reason I brought it up in the context of our last question is it's not just saying, how can we take AI to do this? But if you were to think about the homeowner experience from searching for a home on Redfin all the way through servicing it and you had AI available, what would the ideal experience be like? And it's really interesting to see Rocket with their acquisition strategy to integrate that experience. And that's what I'm excited about. I think there's an opportunity for CEOs and industries like that to have a bold vision of what the future could be. And going back to my point, imperative, not competitive advantage, it is a competitive advantage right now.

32:23

SPEAKER_01

[SPEAKER_00] So if you imagine I haven't tracked the market share of all the big US telcos, but if you look at T-Mobile, Verizon, AT&T, and you tracked it over the past 10 years, you end up with surges in market share growth. The iPhone came out. You ended up with 5G and you end up with these things where, but it's my impression of the industry is you end up with these moments that drive market share and then it ends up at an equilibrium. I think that so it's interesting about it as the iPhone moment for telecommunications companies like SoftBank in Japan. This is the moment where perhaps if you have a competitive equilibrium, you can absorb this technology, use it.

32:35

SPEAKER_00

So if you imagine I haven't tracked the market share of all the big US telcos, but if you look at T-Mobile, Verizon, AT&T, and you tracked it over the past 10 years, you end up with surges in market share growth. The iPhone came out. You ended up with 5G and you end up with these things where, but it's my impression of the industry is you end up with these moments that drive market share and then it ends up at an equilibrium.

32:41

SPEAKER_01

I think that so it's interesting about it as the iPhone moment for telecommunications companies like SoftBank in Japan. This is the moment where perhaps if you have a competitive equilibrium, you can absorb this technology, use it. [SPEAKER_00] Yeah. And you'll have this window where you can actually shuffle the deck. Yeah. It's a technology that shakes the competitive equilibrium. [SPEAKER_00] Yes, exactly right.

32:51

SPEAKER_00

[SPEAKER_01] We'll definitely notice that. So you talked about how coding is such a domain that is suitable to AI because all of the context you're working with exists in the repo. It is in text. It's neatly organized to be executed and read by humans. And so there's a good bounce there. The problem is that customer service agents are not of that character. And so how do you actually put everything into a format where your AI agent can answer it?

32:55

SPEAKER_00

[SPEAKER_01] Yeah. We spend a lot of time thinking about that. One of our engineers called it almost creating a domain specific language for specifying customer experience. What is the mechanism of specifying it? We use this metaphor we call journeys, which is, what is a customer journey end to end? And what does the agent need to be successful in that journey? What tools does it need to access? What information does it need to access? And if you think about the capabilities of the agent like skills and the coding agent, you'll add different capabilities over time as the customer is talking to you. The key thing that's been a breakthrough, which probably isn't surprising to the technologist listening to this, but has been a huge difference between those crappy chat bots of four years ago is the reasoning capabilities. We had one client who had acquired three companies and they had three identity systems, three CRM systems, three of everything.

33:03

SPEAKER_01

And so they had this big IT project where they were going to unify all those systems. But I was thinking, why don't you just have the agent go in all three of them and just think. And they're like, well, what if there's duplicate data? What if the data conflicts? And I was like, that's going to flow. [SPEAKER_00] Sensor fusion. And I was like, well, what does your person, what does a person do? [SPEAKER_00] Well, they kind of think about it.

33:08

SPEAKER_01

And I was like, let's just do that. And that's the interesting thing about these AI agents is they actually, the basic humane, basic reasoning, not superhuman ASI. Turns out to be the huge breakthrough in customer experience. The other interesting thing is the innate knowledge of the LLM. You don't want an AI agent to hallucinate, obviously. But Sonos is one of our clients. And do you have a Sonos speaker at your house? [SPEAKER_00] Or you probably have somebody. I have had, yeah, yeah. If a Sonos speaker ever breaks, it's never the speaker. It's always Wi-Fi. [SPEAKER_00] That's what I've learned. And it's always true of me too, right? There's always some Wi-Fi.

33:21

SPEAKER_01

If you wanted to make an agent to help you with your Sonos speaker, you can give it all the manuals for the speakers, all the technical specifications. You can give it the device telemetry, all the stuff you need. Do you really need to give it the history of Wi-Fi? [SPEAKER_00] Well, it turns out, large language models have encountered every possible Wi-Fi problem. So, why does the Sonos AI work so effectively?

33:30

SPEAKER_01

Well, it knows a lot about Wi-Fi in addition to all the Sonos things. And if you look at any given AI agent, it turns out being trained on all of human knowledge is actually useful as a starting point for a lot of tasks. And I think that's been the big breakthrough. So, how do you give it all of its knowledge?

33:32

SPEAKER_01

[SPEAKER_00] Well, first, we've built, I think, the best platform in the market to do so, where you can really narrow the guardrails for regulated conversations, widen them for less regulated conversations. But the fact it starts with knowledge of obscure Wi-Fi idiosyncrasies turns out to be the greatest breakthrough of all time. Have you had the opposite problem with a customer whose problem domains mostly don't exist on the public internet? You know, it's like, we provide the drill bits used in deep-sea oil drilling. It turns out there's nothing on Reddit about that.

33:36

SPEAKER_01

[SPEAKER_00] Yeah, so 100%. And we work with this medical device company, and it's a deep cut of human knowledge. And you can train it all on that. In fact, we do a lot. One of the things you want to be really careful about if you have a really well-known brand, and we work with, I want to say, a third of our clients have over 10 billion in revenue. Over half have over a billion revenue. So, most of our clients are actually quite well-known. So, one of the challenges when you're offering either sales or service or customer experience to a really well-known brand is it's harder to ground it. Yeah.

33:39

SPEAKER_01

[SPEAKER_00] It's actually easier when the internet has never heard of you and you want to make a well-grounded agent. It's actually pretty easy because there's no temptation from the LLMs to go off script. So, actually, I would say— They know what they don't know. [SPEAKER_00] Ironically, the harder challenge is when it's a very well-known brand.

33:45

SPEAKER_00

[SPEAKER_01] So, most of our clients are actually quite well-known. [SPEAKER_01] So, one of the challenges when you're offering either sales or service or customer experience to a really well-known brand is it's harder to ground it. [SPEAKER_01] Yeah. [SPEAKER_01] It's actually easier when the internet has never heard of you and you want to make a well-grounded agent. [SPEAKER_01] It's actually pretty easy because there's no temptation from the LLMs to go off script.

34:06

[SPEAKER_00] [SPEAKER_01] So, actually, I would say- [SPEAKER_00] [SPEAKER_01] They know what they don't know. [SPEAKER_00] Ironically, the harder challenge is when it's a very well-known brand. [SPEAKER_00] It's like, no, I got this. [SPEAKER_00] [SPEAKER_01] I'm like, no, you don't. [SPEAKER_01] You got to go look it up. [SPEAKER_01] That's actually a harder problem. [SPEAKER_01] And so, how do you force the LLMs, mechanically, how do you force them to not answer off the top of their head but actually look it up?

34:56

SPEAKER_01

So, we use what we call a constellation of models. [SPEAKER_00] So, our platform, we call it Agent Studio. You essentially configure the goals and guardrails of a process. And goals and guardrails, not a sequence of steps because you want agency.

35:11

SPEAKER_00

[SPEAKER_01] You want guardrails around it. [SPEAKER_01] And within that, we'll use reasoning.

35:14

SPEAKER_01

But we use supervisor models to actually inspect that reasoning. And so, if you were an AI agent in Sierra and you decided to go off script, I got this. I don't know, what would end up happening is a supervisor agent would observe your reasoning, say, I think John should have actually looked up the policy here and send it back with notes. And say, actually, you're not allowed to make that decision, here's the reasons why, go redo that decision. [SPEAKER_00] [SPEAKER_01] It's a really effective technique.

35:36

The way I think about it, which is a little simplistic, but I think basically right, if you imagine a reasoning system is right 90% of the time but has some either guardrail malfunction or hallucination 10% of the time, it's obviously better than that. [SPEAKER_01] And then you have a supervisor that's right 90% of the time. [SPEAKER_01] If you chain them together, you get 99% effectiveness.

35:45

SPEAKER_01

And so, that methodology of layering reasoning and intelligence has been really effective. And in general, it makes sense. [SPEAKER_00] [SPEAKER_01] You're basically layering compute. You're layering reasoning on top of it. What's neat about it, though, is we can abstract that complexity from our clients.

35:54

SPEAKER_00

[SPEAKER_01] So, they're expressing the goals and guardrails. [SPEAKER_01] And we have all these evals and tests and all these other things.

36:02

SPEAKER_01

We can find ways to make it more and more and more robust over time.

36:07

SPEAKER_00

[SPEAKER_01] But it doesn't require you to prompt engineer or write in all caps or whatever the hacks that people use to get these things to be conformant. [SPEAKER_01] And you started in 22, 23?

36:09

SPEAKER_01

We launched the company on February 13th, two years ago. So a little world like our- [SPEAKER_00] Oh, 24. [SPEAKER_00] Yeah.

36:17

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] So our two-year birthday was a couple weeks ago. [SPEAKER_01] Because what I was thinking as you were saying that is, did you co-evolve chain of thought and RL and some of these things that are now in the models? [SPEAKER_01] But did you have to build your own version of them before they were in the models? [SPEAKER_01] So yes. [SPEAKER_01] And also talk about the weird part about building a product right now and a company right now. [SPEAKER_01] Because so much of what we write we plan to throw out later.

36:47

SPEAKER_01

Yes. And this is a very weird way to build a company. [SPEAKER_00] So Google's chain of thought paper, which preceded 01 and doing reinforcement learning chains of thought, was out roughly when we started the company. It was an earlier paper and effectively provided a substantive basis of why asking a model to explain its reasoning step-by-step produced more robustness. So we used chain of thought all the time. [SPEAKER_00] And it was a methodology we used. [SPEAKER_00] And then OpenAI came up with the idea of doing reinforcement learning on those chains of thought, which is where 01 came from. And then most labs are doing that now. So we'd throw out things all the time.

37:35

SPEAKER_01

You'd do it and you're like, okay, the model just does this for us now. We work with a lot of financial services firms. [SPEAKER_00] We work with one bank that has a large Hong Kong business and they speak Cantonese. And okay, well, we need really good Cantonese voice support.

37:58

SPEAKER_00

[SPEAKER_01] And it turns out that that's really hard and there's not an obvious model that does that.

38:12

SPEAKER_01

So we spend all this time evaluating all these models. Yes. What certainty would you ascribe to every voice model supporting Cantonese within three years? A hundred percent? Ninety-nine percent?

38:39

SPEAKER_00

[SPEAKER_01] Pretty close.

38:42

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] Yeah. So we did all this work. [SPEAKER_00] [SPEAKER_01] And in fact, I think we have the best Cantonese support on the market. [SPEAKER_00] [SPEAKER_01] Great for us. [SPEAKER_00] [SPEAKER_01] It's a huge selling point. And it's a technology that will certainly be commoditized in three years. So a lot of what we think about is going from technology innovation now. [SPEAKER_00] [SPEAKER_01] I think a large part of why we work with the largest companies in the world is because our technology works. [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] In three years, those same clients will work with us because we have the best product.

39:09

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And I think if you look at the early marketing for early SaaS companies, they'll explain why having multiple tenants in the same database is safe. [SPEAKER_00] [SPEAKER_01] And that was a huge part of their marketing. [SPEAKER_00] [SPEAKER_01] Nowadays, if you came and marketed your product that way, people are like, what are you talking about? [SPEAKER_00] [SPEAKER_01] I don't care what database Stripe uses. [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] In three years, those same clients will work with us because we have the best product.

39:43

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And I think if you look at the early marketing for early SaaS companies, they'll explain why having multiple tenants in the same database is safe. [SPEAKER_00] [SPEAKER_01] And that was a huge part of their marketing. [SPEAKER_00] [SPEAKER_01] Nowadays, if you came and marketed your product that way, people are like, what are you talking about? [SPEAKER_00] [SPEAKER_01] I don't care what database Stripe uses. [SPEAKER_00] So I think we're just at this period where the technology is so immature. It's a very technology forward conversation just because people are figuring it out.

40:08

SPEAKER_01

Just when Netscape's business was monetized through a web server back 100 years ago. [SPEAKER_00] And it will evolve from being a technology forward conversation to a product forward conversation.

40:26

SPEAKER_00

[SPEAKER_01] So the interesting part about building an applied AI company is you can't have the luxury of waiting for all the models to catch up with your aspirations.

40:31

SPEAKER_01

But you know they will. [SPEAKER_00] But you know they will. [SPEAKER_00] Yes. So you have to have the best technology and have to be comfortable with throwing it out. Yes. And it's a real momentum and pace of innovation game rather than thinking of this as precious intellectual property, if that makes any sense.

40:44

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] It absolutely does.

40:45

SPEAKER_01

But isn't this organizationally hard where if I'm the head of Cantonese language at Sierra, my incentive and then not disingenuously so, I'll notice all the corner cases where the models aren't that good in Cantonese. And obviously we saw this in prior tech waves, right? Where the cloud adoption laggards were companies that had their own on-prem stuff and they had a million reasons. Yeah. Half real, half fake as to why cloud did not suit their business purposes. But how do you avoid getting stuck in this mode of thinking where, well, their chain of thought doesn't do what we need is the classic thing you would hear from someone within the organization. It's a huge shift.

41:02

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] I mean, going back to the first thing we were talking about, it's hard for me to not care about the elegance of the source code, which I think is an impediment to my fully realizing being a software engineer in this new world. [SPEAKER_00] [SPEAKER_01] I think teams that start to treat the code that they wrote as precious that has been obviated by a general purpose AI model will fundamentally fall behind. Public markets deem the software industry 20, 30% less valuable than they did three months back. A day ago. Yeah, exactly. Very recently.

41:18

SPEAKER_01

And the two sides of the debate are one, the valuations were based on what the businesses will do in 2030 or 2035, far in the future.

41:19

SPEAKER_00

[SPEAKER_01] And there's much more uncertainty there.

41:21

SPEAKER_01

And so this is deserved.

41:22

[SPEAKER_01] And the counter argument is that it's still not the case that agentic software production is really going to build you a Workday. [SPEAKER_01] Indeed, Anthropic just installed Workday very famously. [SPEAKER_01] Where do you net out on this: is this a rational response or not? [SPEAKER_01] I think it's rational, but I think it's a bit overblown at the same time. So I think it's rational in the sense that there probably hasn't been more uncertainty in this market ever.

41:36

SPEAKER_00

[SPEAKER_01] Yes, yes. [SPEAKER_01] And so unless you have a strong thesis about an individual company, my guess is will these companies be less valuable 10 years from now than now? [SPEAKER_01] I think the answer is probably yes. [SPEAKER_01] Will that be true for every individual company? [SPEAKER_01] Indeed, Anthropic just installed Workday very famously. [SPEAKER_01] Where do you net out on this: is this a rational response or not? [SPEAKER_01] I think it's rational, but I think it's a bit overblown at the same time.

42:05

SPEAKER_01

So I think it's rational just in the sense that there probably hasn't been more uncertainty in this market ever. Yes, yes. And so unless you have a strong thesis about an individual company, my guess is will these companies be less valuable 10 years from now than now? I think the answer is probably yes. Will that be true for every individual company? I don't think that's true. And so if you're just thinking about a portfolio of investments, I think it's an indictment of the sector more than it is an indictment of an individual company. [SPEAKER_00] I don't know if the value of these platforms was who could code in a weekend ever. Not that we knew what coding was.

42:44

SPEAKER_01

My point is, everyone who's ever built a software as a service application has had a Hacker News comment of I could have coded this in a weekend. Every single one. [SPEAKER_00] [SPEAKER_01] Famously Dropbox, I'm sure you have as well. [SPEAKER_00] [SPEAKER_01] Every single product I've ever made. [SPEAKER_00] [SPEAKER_01] It just happens. It's a rite of passage. In fact, if no one said that on your product, I'm sorry. It's not relevant. Yeah, you're not relevant. That's not interesting. And obviously most of those comments were incorrect.

43:21

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] But if you think about all the work you've done in compliance or the relationships you have with large financial services institutions, the work you do on fraud, the things under the surface that aren't the forms and fields in the web browser are actually incredibly valuable. [SPEAKER_01] If you think about a large software company, they'll have thousands of quota carrying account executives or represent sales capacity, which is a channel.

43:25

SPEAKER_01

Yeah. [SPEAKER_00] [SPEAKER_01] And distribution turns out to be a very important part of software. And there's social proof. There's the old saying, no one gets fired for buying IBM, which few people say right now. [SPEAKER_00] Though IBM is actually doing really well under Arvind.

43:41

SPEAKER_00

You want to be maybe the first health care insurance company to adopt something.

43:45

SPEAKER_01

[SPEAKER_00] There's another health care insurer who says, I want to be the fifth. You know, I want other people to prove it. [SPEAKER_00] There's all these network effects around these businesses and scale and moats. I think the big risk is, where is value in the software industry years from now. [SPEAKER_00] One risk is that more people will build than they do now versus buy because the marginal cost of writing software goes down. [SPEAKER_00] I think that'll be true for some software, particularly developer platforms and things that are already being consumed in purchases by other engineers.

44:03

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] Little libraries or things that already were part of the build versus buy calculus. It shifts the balance of power. Absolutely. Yeah. [SPEAKER_00] The other part of it is systems of record. [SPEAKER_00] One risk is that more people will build than they do now versus buy because the marginal cost of writing software goes down. [SPEAKER_00] I think that'll be true for some software, particularly developer platforms and things that are already being consumed in purchases by other engineers. [SPEAKER_00] [SPEAKER_01] Little libraries or things that already were part of the build versus buy calculus. It shifts the balance of power. Absolutely. Yeah.

45:03

SPEAKER_01

[SPEAKER_00] The other part of it is systems of record. [SPEAKER_00] So I think these systems of record have always been the gravitational center of their relative solar systems. [SPEAKER_00] And it roughly breaks down by department. [SPEAKER_00] So ERP systems are associated with the finance department and SAP and Oracle and Workday have ERP systems. [SPEAKER_00] And you have Adobe in the marketing department. [SPEAKER_00] And you had Salesforce in the sales department. [SPEAKER_00] And you had ServiceNow in the IT department. [SPEAKER_00] And everything rotated around them.

45:38

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] And why? [SPEAKER_01] Well, first, their database was the system of records. [SPEAKER_01] So every application that wanted to interact with the data had to. You essentially collect taxes from your ecosystem.

45:42

SPEAKER_01

And then similarly allowed each of those systems of record company to essentially have revenue expansion opportunities to go to adjacent areas where they're all sold to the same buyer. And the thing that's really interesting is AI agents are actually performing valuable labor. Is the database and the system of record, does that continue to be the gravitational center of each of those workflows? So I'll just take marketing as an example. The database of your customers that you use to drive sending out an email blast on Black Friday has some value.

46:03

SPEAKER_01

But if you had an AI agent that drove way higher, more leads for your sales team from that marketing blast, you probably value that more than the system of record itself. Similarly, if you imagine a CRM system and you think about the AI agent that's carving your territories if no one ever logs in to actually do it manually. All of those things have a lot of value and a lot more value than, relatively speaking, they did because they're actually performing the action.

46:07

SPEAKER_00

[SPEAKER_01] And so the real question to me is, does it upend this, something that's been true for 30 years, which is all the values in these systems of record.

46:15

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And the way I think about it is agents are to some degree a system of record of a process of generating a lead or auditing your financials or reviewing a contract or whatever it might be. And I don't think we've ever had a piece of software that.

46:29

SPEAKER_00

[SPEAKER_01] And will those encoded, well-optimized processes start to have more value than the databases? [SPEAKER_01] I don't know that's the case.

46:41

SPEAKER_01

For example, if your ERP system is your company's ledger, that'll have a lot of value. [SPEAKER_00] But I wonder for all these others, and my theory is the closer you get to literally the database is the value, i.e. a ledger, the more durable it is. [SPEAKER_00] The closer you get to being a system of engagement, the less durable it is. [SPEAKER_00] That's a very interesting framing. Yeah.

46:52

SPEAKER_00

[SPEAKER_01] And it gets back to the point you were making about the company that was looking to standardize and not have three different ERPs and all that.

46:57

SPEAKER_01

And you're, well, why, just try not doing that. [SPEAKER_00] [SPEAKER_01] And I think maybe the consumer example of this is, I think people have probably had the experience of pasting data into an LLM to do something with it. And the formatting is all messed up and the tabs and spaces don't come through and everything. So it doesn't matter. Yeah. LLM doesn't care. You can just paste through whatever and it'll work with it. And so this idea that, as you say, if the system of record is important because it's your general ledger and it matters to the auditors, that's one thing.

47:17

SPEAKER_01

But if it was a system of record in this all your data in one place way, and because it was easier to build incremental software atop it, maybe that advantage is going away because the agents are fine plucking data from 10 different places. And that's roughly my view, but the bull case, I always mix up bear and bull. I need to spend more time on Wall Street. The bull case though is, I think all these companies have a right to win. You can just pay through whatever and it'll work with it. And so this idea that, as you say, if the system of record is important because it's your general ledger and it matters to the auditors, that's one thing.

47:29

SPEAKER_01

But if it was a system of record in this all your data in one place way, and because it was easier to build incremental software atop it, maybe that advantage is going away because the agents are fine plucking data from 10 different places. And that's roughly my view, but the bull case, I always mix up bear and bull. I need to spend more time on Wall Street. The bull case though is, I think all these companies have a right to win. They're all big, they still have sales capacity, they have all these advantages, but it's a race. [SPEAKER_00] How fast will smaller companies build differentiated scaled businesses before the incumbents grow into this new world?

47:58

SPEAKER_01

[SPEAKER_00] But for a wide variety of well-documented reasons, disrupting their own business model is harder. [SPEAKER_00] But I think your ask is, is it irrational? [SPEAKER_00] I don't think it's irrational.

48:07

SPEAKER_00

[SPEAKER_01] I think there's just more uncertainty now than there's ever been.

48:12

SPEAKER_01

And I think that's what markets are telling you—there's a lot of uncertainty and that's why you see people recede from the whole category. I feel there's also a totally separate thing playing out here where for a long time certain companies were criticized for not taking profitability that seriously.

48:25

SPEAKER_00

[SPEAKER_01] And at some level, there's just a return to normal valuation levels on a fully loaded stock-based comp basis.

48:30

SPEAKER_01

That's independent of the AI thesis, but maybe just some return to more normal valuations on a fully loaded gap basis. Well, essentially, if you look at a traditional software as a service company, the way most people model it is you have annual recurring revenue, which is an annuity.

48:38

SPEAKER_00

[SPEAKER_01] Yep. [SPEAKER_01] And it should throw off that much cash every year.

48:39

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] Then you have attrition, which is subtracting from the annuity. And then you have net new ARR, which is adding to the annuity. Your salespeople sell software to add to the ARR.

48:52

SPEAKER_00

[SPEAKER_01] You typically have an account management team or customer success team to keep churn down.

48:53

SPEAKER_01

And you grow that annuity and you grow your headcount often just a little bit ahead of that annuity because you need to grow a new business. [SPEAKER_01] And if that annuity is not an annuity. Yeah. Then that math really changes. It really changes. Yep. And it should throw off that much cash every year. [SPEAKER_00] [SPEAKER_01] Then you have attrition, which is subtracting from the annuity. [SPEAKER_01] And then you have net new ARR, which is adding to the annuity. Your salespeople sell software to add to the ARR. You typically have an account management team or customer success team to keep churn down.

50:08

SPEAKER_01

And you grow that annuity and you grow your headcount often just a little bit ahead of that annuity because you need to grow new business. And if that annuity is not an annuity. Yeah. Then that math really changes. It really changes. And so, because the whole idea of software as a service is you can slow down hiring and you become very profitable because the annuity starts throwing off cash. And so, I think that's been the thesis of every private equity firm who acquires slow growth software as a service companies. If you don't assume that revenue is going to be there two or three years from now, your discounted cash flow analysis looks pretty different.

50:44

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And I don't think it's actually quite so dire in the timeframes that people think. But again, if you're asking what markets are, there's all those great quotes about voting machines. Voting machines. Yeah, I get it. There's probably safer sectors to invest in. But I don't think it's an indictment of individual companies. [SPEAKER_00] [SPEAKER_01] That's my point. [SPEAKER_00] I actually think when we first met, I doubt either of us had an extremely positive view of the future of Microsoft. [SPEAKER_00] At the time it felt like a previous generation company.

51:22

SPEAKER_01

[SPEAKER_00] Now you look at Azure, their open relationship, all these things—what an impressive turnaround.

51:31

SPEAKER_00

So, I think any one of these companies could do it.

51:37

SPEAKER_01

[SPEAKER_00] I think it's just more of an indictment of the market. [SPEAKER_00] Yes, yes. [SPEAKER_00] I have a lot more questions. [SPEAKER_00] Would you like to know the Guinness? Sure. [SPEAKER_00] Brett has been through a few platform shifts. [SPEAKER_00] And one thing he's been pretty consistent about is being mindful of the external forces that are shaping the ecosystem you're in. [SPEAKER_00] [SPEAKER_01] He talks a lot about building with the broader wave of AI agents in mind. [SPEAKER_00] [SPEAKER_01] Stripe Sessions is our way of helping builders see that wave up close.

51:57

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] What's changing in the internet economy, what's actually working in production, and what the next era of software looks like when agents are running real commerce workflows. [SPEAKER_00] [SPEAKER_01] It's not the usual conference fluff.

52:02

SPEAKER_00

[SPEAKER_01] It's insights into what the fastest moving companies are actually up to.

52:17

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] So, if you want to experience the next chapter of the internet economy firsthand, join us at Stripe Sessions this April. [SPEAKER_00] [SPEAKER_01] Use the code CheekyPint for 50% off a conference pass at sessions.stripe.com. [SPEAKER_00] [SPEAKER_01] You talked about business models. [SPEAKER_00] [SPEAKER_01] Are you guys usage-based or how are you innovating on the business model front, or are you? [SPEAKER_00] We are trying to.

52:38

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] So, we do outcomes-based pricing. [SPEAKER_01] Yep. [SPEAKER_01] So, for a customer service context, that means if the AI agent resolves the case with no human intervention, there's a pre-negotiated rate for that. [SPEAKER_01] And if we do have to escalate to a person, that's free for sales; it would be a sales commission. [SPEAKER_01] Yep. [SPEAKER_01] And wherever possible, there's a way to align our interests with our clients. [SPEAKER_01] Yeah. [SPEAKER_01] We choose it.

53:26

SPEAKER_01

And I'm a huge believer in this. [SPEAKER_01] I think the analogy of going from impression-based ads to CPC ads is apt. Yes. [SPEAKER_00] I don't think any ad platform thinks, "Man, think of all the impressions we're giving away for free."

53:34

SPEAKER_00

[SPEAKER_01] Because when you charge for something closer to business value, it's actually more valuable. [SPEAKER_01] It's more efficient. [SPEAKER_01] It's a lot more efficient. [SPEAKER_01] Yeah. [SPEAKER_01] And I think the idea, if an agent's outcome is measurable, it's a really compelling way to, both for clients obviously, because it's aligned with their business. But it's also quite disruptive because most legacy software companies are not necessarily equipped to do it for a variety of reasons I'm happy to go into. But it's just a very disruptive model. [SPEAKER_01] Yeah. [SPEAKER_01] There's a few lenses you can have on it.

54:05

SPEAKER_00

[SPEAKER_01] One is that you get more alignment.

54:07

SPEAKER_01

Usage-based is more aligned than other ways of charging. And as you say, it's more efficient because you're incentivized to drive the right outcomes. People also make analogies to, it's almost more correct for the labor substitution dynamics that you get. Or just because you have real inference costs, you're going to have to do a usage-based model. Do those factor in at all? [SPEAKER_00] But it's just a very disruptive model. Yeah. There's a few lenses you can have on it.

55:12

SPEAKER_01

One is that you get more alignment. Usage-based is more aligned than other ways of charging. And as you say, it's more efficient because you're incentivized to drive the right outcomes. People also make analogies to it's almost more correct for the labor substitution dynamics that you get. [SPEAKER_00] [SPEAKER_01] Or because you have real inference costs, you're going to have to do a usage-based model. [SPEAKER_00] [SPEAKER_01] Do those factor in at all? [SPEAKER_00] Or would it not be possible to do a fixed price contract because— [SPEAKER_00] I would actually argue outcomes-based is pretty different than usage-based.

56:12

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] Okay. [SPEAKER_01] [SPEAKER_00] Think of it this way. [SPEAKER_01] [SPEAKER_00] If you have an AI agent that is making sales for Stripe to small businesses, and I told you I will sell one-tenth the number of new Stripe GMV, however you'd value that.

56:31

SPEAKER_01

[SPEAKER_00] But I'll use one-hundredth of the tokens, you probably wouldn't care. [SPEAKER_00] You care about the value to your top-line of your business. [SPEAKER_00] I would argue there's not a strong correlation between token usage or utilization and value. [SPEAKER_00] There may be, but there's not always. [SPEAKER_00] There was that infamous website, I think it was called Folklore, where that Apple engineer used to put all this Apple Folklore. [SPEAKER_00] Folklore.org, yeah. [SPEAKER_00] Folklore.org, yeah, I love it. [SPEAKER_00] It's a fun site to go to if you're an engineer. [SPEAKER_00] You care about the value to your top-line of your business.

57:01

SPEAKER_01

[SPEAKER_00] I would argue there's not a strong correlation between token usage or utilization and value. [SPEAKER_00] There may be, but there's not always. [SPEAKER_00] There was that infamous website, I think it was called Folklore, where that Apple engineer used to put all this Apple Folklore. [SPEAKER_00] Folklore.org, yeah. [SPEAKER_00] Folklore.org, yeah, I love it. [SPEAKER_00] It's a fun site to go to if you're an engineer.

57:44

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] But there was a story about some new manager asking for lines of code every day, and one of the engineers wrote a negative number as a way of saying something to the manager because he refactored a code base or whatever. [SPEAKER_01] [SPEAKER_00] I think that is the essence of why tokens are not correlated with value. [SPEAKER_01] [SPEAKER_00] They may be, but the idea that they definitely are, I don't think stands to reason. [SPEAKER_01] And so usage-based is charging for storage or something.

58:04

SPEAKER_01

Outcomes-based is what business outcome is this agent designed to produce and did it produce it effectively. That is really aligning because it creates this whole vertical alignment. So as a company, reducing your token utilization for the same outcomes is your problem, not your customers. And that's a great incentive to drive more efficiencies over time. It means that to grow your relationship with the client, you actually have to make your product better. [SPEAKER_01] Yeah. Not just theoretically better, but actually better.

58:58

SPEAKER_01

How do you have usage-based or outcomes-based when you move beyond customer service where there's a clear "was this resolved or not" to product usage where people are shopping and they didn't buy a house there. But people mostly don't buy a house on most website visits, but it was a successful visit. So it's the right question and there's not a great way to do it for every type of agent right now. And so you can fall back to usage-based, which is fine. Yeah, yeah. But over time, wouldn't it be interesting? I think AI agents should have memory. [SPEAKER_00] I think AI agents should drive relationships, not conversations. [SPEAKER_00] [SPEAKER_01] Yeah.

59:31

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And it would be really interesting to say, could we make an AI agent that actually drives home ownership over time? [SPEAKER_00] [SPEAKER_01] I think that's actually hard. [SPEAKER_00] [SPEAKER_01] Yeah, yeah, yeah. [SPEAKER_00] [SPEAKER_01] But it's not— [SPEAKER_00] [SPEAKER_01] Is an AI agent that you have a territory? [SPEAKER_00] [SPEAKER_01] I think so. [SPEAKER_00] I mean, even because it's hard today and we're a pragmatic company, I think it's the right thing to ask, because that's fundamentally the value the software is designed to produce. [SPEAKER_00] And so I think it's a values aligning thing.

1:00:09

SPEAKER_01

[SPEAKER_00] It also changes the dynamics of a software company's relationship to its partners, to its clients, because if you go back ancient history four years ago, there was a stark separation between software and implementation and usage. [SPEAKER_00] It was the client's accountability to use the product well. [SPEAKER_00] Yeah. [SPEAKER_00] It was either your IT team or a systems integrator's responsibility to implement the software and the job of the software company was to make it and throw it over the wall. [SPEAKER_00] Obviously, it's not exactly that, but that was the market we were in.

1:00:18

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] And everyone had good intentions, but success has a thousand fathers, failures are orphans. [SPEAKER_01] [SPEAKER_00] And when the software didn't go well, everyone was blaming everyone else. The client was like, I'm using it just fine. It was implemented poorly. [SPEAKER_01] The person implementing it was like, no, the platform's broken. [SPEAKER_01] The platform people would say, and it was everyone pointing at everyone else. [SPEAKER_01] What's nice about outcomes-based is, whether or not the client sets it up, you become more accountable to help them be successful. [SPEAKER_01] Because until they do, they can't use it.

1:00:43

SPEAKER_00

[SPEAKER_01] If there is some last mile of implementation, it creates a strong incentive for the software company to have skin in the game to help you navigate that last mile. [SPEAKER_01] I think so many problems in the software industry are due to that lack of accountability.

1:00:46

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] If you talk to any company that's ever implemented an ERP system, it's a multi-year process.

1:00:50

SPEAKER_00

[SPEAKER_01] It's invading Russia. [SPEAKER_01] Yeah. [SPEAKER_01] And you don't even remember why you're doing it midway through.

1:00:58

SPEAKER_01

[SPEAKER_00] You've gone through two CFOs and three CIOs by the time it's done. And we're okay with that. That's just the way software works. And so my view is, I think AdWords sort of changed the advertising industry on the internet and drove it. And I think you're going to pay for mobile app installs and now directly pay for outcomes. I think it's a really positive step forward. It's not going to be possible for everything. You have to have pragmatism.

1:01:32

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] But I think it's the right way to have a partnership. You should share in the outcomes.

1:01:39

SPEAKER_00

[SPEAKER_01] You want to wire the company to be thinking in this outcomes-based way. [SPEAKER_01] In your main customer service stuff, you can do that in other ways you might not be able to yet, but you want people to be spring-loaded to be thinking that way. [SPEAKER_01] That's right.

1:01:43

SPEAKER_01

And if the whole company is incentivized towards outcomes, it's a way better partner to work with because of it. [SPEAKER_00] I mean, we find this at Stripe where we have outcomes-based pricing. [SPEAKER_00] You've always had outcomes. Exactly. [SPEAKER_00] [SPEAKER_01] Yeah, it's transactional, but we find there's a lot of uplift we can get on just getting people more revenue and finding ways to. We're sometimes hammering customers where it's like, you should be accepting local payment methods for internationalization. [SPEAKER_00] [SPEAKER_01] Yeah. That's right.

1:02:02

SPEAKER_01

And if the whole company is incentivized towards outcomes, it's a way better partner to work with because of it. [SPEAKER_00] We find this at Stripe where again, we have outcome based pricing. [SPEAKER_00] You've always had outcomes.

1:02:11

SPEAKER_00

[SPEAKER_01] Exactly.

1:02:16

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] Yeah, it's transactional, but we find there's a lot of uplift we can get on just getting people more revenue and finding ways to, we're sometimes hammering customers where it's like, you should be accepting local payment methods for internationalization. [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] Or you're crazy not to be turning on this feature, but we really feel it because we have the same incentive as the customer. [SPEAKER_00] [SPEAKER_01] It's like, this will be revenue maximizing for both of us. [SPEAKER_00] [SPEAKER_01] I'm going to ask a very AGI brained question. [SPEAKER_00] [SPEAKER_01] I just can't resist.

1:02:35

SPEAKER_00

[SPEAKER_00] [SPEAKER_01] I'm glad we're in a second. [SPEAKER_01] Exactly. [SPEAKER_01] Now that we get to it, which is you described building stuff that you know you're going to throw away because the model capabilities will get there.

1:02:46

SPEAKER_01

[SPEAKER_00] And you're like, occasionally they are developing capabilities that you developed yourself. [SPEAKER_00] [SPEAKER_01] Isn't Sierra itself a shortage? [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] Sorry, I said I couldn't resist. [SPEAKER_00] [SPEAKER_01] No, it's the right question. [SPEAKER_00] [SPEAKER_01] You know, the short answer is I don't know. [SPEAKER_00] [SPEAKER_01] The fog of war in the software industry is pretty thick right now. [SPEAKER_00] [SPEAKER_01] Yes. I really believe in the applied AI market though.

1:03:15

SPEAKER_00

[SPEAKER_01] I think most companies don't want to buy models or buy software. [SPEAKER_01] They want to buy solutions to their problem. [SPEAKER_01] And if you just go back to the cloud industry, why doesn't Amazon and Microsoft do everything for everyone? [SPEAKER_01] There's not really a reason by somewhat similar logic, why should any software as a service company exist when you have bigger scale, all this technology? [SPEAKER_01] In theory, they could just develop all the software. [SPEAKER_01] And actually, many of them have tried. [SPEAKER_01] There's actually competitors to Salesforce and almost all of the above.

1:03:48

SPEAKER_00

[SPEAKER_01] I think there's so much nuance in how these companies align themselves with different departments at these companies, solve their very unique problems in very specific ways. [SPEAKER_01] [SPEAKER_00] That is a mix of product, not technology, but product, go to market. [SPEAKER_01] It's an ecosystem around it. [SPEAKER_01] And I think a lot of that still exists because I'm not sure coding the software was necessarily the hard part. [SPEAKER_01] And then similarly, I actually think especially in enterprise software, how you engage with your clients really matters.

1:04:09

SPEAKER_00

[SPEAKER_01] And I think it turns out that GPT-5 and Claude, whatever version it's on right now, or Opus, excuse me, is sold to a different buyer than the CFO or the chief customer officer or the chief digital officer. [SPEAKER_01] And that seems small, but it's actually big. [SPEAKER_01] And so I think you tend to see software companies orient around individual buyers within companies. [SPEAKER_01] You tend to see consolidation around departments and around buyers. [SPEAKER_01] It's possible that you can go beyond those lines, but it hasn't happened traditionally.

1:04:35

SPEAKER_01

And I think the reason for it is most business users want actual solutions to their problems and they want a company that serves their unique problems in a very specific and bespoke way. So I actually am extremely bullish on applied AI. I actually think we could accelerate, I'll make one statement, which is I think if we paused model development, we'd still have trillions of dollars of economic value. [SPEAKER_00] I totally agree. [SPEAKER_00] That have yet to be realized.

1:05:02

SPEAKER_01

[SPEAKER_00] And I think if we had a mature applied AI market where the CFO could go buy that agent to onboard new supply chain vendors that just worked, we could actually accelerate that trillions of dollars of economic value. [SPEAKER_00] [SPEAKER_01] So I think not only am I somewhat skeptical that there will only be two companies in the world, I actually think one of the main things impeding adoption of AI is the lack of existence of all those other companies.

1:05:07

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And so many of the startups, particularly around here in San Francisco, are basically doing relatively rote tools around the AI rather than actually building agents for business processes that are boring, but important and valuable. [SPEAKER_00] So I'm really bullish on it. [SPEAKER_00] Yeah, yeah.

1:05:34

SPEAKER_01

[SPEAKER_00] And I guess you help companies ensure that they can always have access to the latest models, which sounds like a minor thing, but the leading model is always changing. [SPEAKER_00] And so that's not trivial. [SPEAKER_00] I agree. [SPEAKER_00] I'm not sure how much of a long term value it is. [SPEAKER_00] I think it is.

1:05:49

SPEAKER_00

[SPEAKER_01] Up to this point, the race has been led by a matter of months, right?

1:05:52

SPEAKER_01

[SPEAKER_00] Well, every single month, there's a new frontier model and your customer experience doesn't change that frequently. So you're absolutely right. [SPEAKER_00] But I also think there's just a big product right. [SPEAKER_00] Like our clients use it to optimize their sales. And that is a product, not a technology. [SPEAKER_00] And it's very particular to the workflows of people building customer experience teams, building sales teams. And that's what we're focused on. And I think those departments deserve purpose built software.

1:06:07

SPEAKER_00

[SPEAKER_01] And I think there will be enduring value there. [SPEAKER_01] But it's interesting. [SPEAKER_01] It's the right question to ask. [SPEAKER_01] I don't think we've ever lived in a world where production of software was easy. [SPEAKER_01] And software engineering was the most scarce asset in a company. [SPEAKER_01] And now it's the most plentiful. [SPEAKER_01] And I don't think we've ever lived in that world. [SPEAKER_01] Yes. [SPEAKER_01] Yes. Well, that gets to one of the biggest conundrums in Silicon Valley right now, which is what will the shape of the world of AI productivity be.

1:07:13

SPEAKER_00

[SPEAKER_01] And I think there's a strong sense that AI has gotten really good and it should change the composition of companies and should change the hiring plans somewhat.

1:07:14

SPEAKER_01

And you've seen this in some corners. Block announced their 45%, 50% AI layoff yesterday. And you have some companies not growing as quickly. At the same time, in coding, you see a lot of AI benefits. So you can argue that either way, right? Right. You can say engineers have gotten much more productive. [SPEAKER_00] [SPEAKER_01] Therefore, we should hire fewer engineers. [SPEAKER_00] [SPEAKER_01] Or you could say engineers have gotten much more productive. The ROI on a single engineer is way higher. We now have super engineers that we can hire.

1:07:59

SPEAKER_00

[SPEAKER_01] Therefore, we should hire way more of them.

1:08:03

SPEAKER_01

And you've seen this in some corners. Block announced their 45%, 50% AI layoff yesterday. And you have some companies not growing as quickly. At the same time, in coding, you see a lot of AI benefits. So you can argue that either way, right? You can say engineers have gotten much more productive. Therefore, we should hire fewer engineers. Or you could say engineers have gotten much more productive. The ROI on a single engineer is way higher. We now have super engineers that we can hire. Therefore, we should hire way more of them. And because there isn't a fixed amount of stuff for Stripe or any other company to do.

1:08:40

SPEAKER_01

And then the AI productivity story in other roles is just a bit less clear because as we've discussed, AI is uniquely well suited to coding. So what do you make of how the AI productivity shows up?

1:08:48

SPEAKER_00

[SPEAKER_01] I feel every company in Silicon Valley is trying to figure this out right now. [SPEAKER_01] Well, first, I'll go back to why I believe in applied AI.

1:08:52

SPEAKER_01

I think the atomic unit of productivity in AI is a process, not a person. I don't think AI - I don't know if you have an assistant, but if you do, he or she might help you prepare for a podcast, might help you prepare for a meeting. He or she might also get you a cup of coffee. [SPEAKER_00] So no matter AGI, short of robotics, will get you a cup of coffee. [SPEAKER_00] So I think it's wrong to think about AI as replacing people in addition to being inhumane. [SPEAKER_00] It's nonsensical because AI operates in the world of digital technologies.

1:09:17

SPEAKER_01

And I think if you go to an example of even a mundane process in your business, onboarding a new supplier, think about all the departments and people involved in that. There's a legal department to do a contract. There's a finance department, procurement to negotiate the relationship. You probably have IT that's involved to onboard them into your core systems. [SPEAKER_00] And then there's usually a business response right now.

1:09:28

SPEAKER_00

Fairly mundane, happens all the time.

1:09:32

SPEAKER_01

If you tracked what is the median amount of time it takes to onboard a new supplier and it was 17 days, just for argument's sake. I bet you could say as a CEO of a company, I want to use AI to optimize that process and make it 17 hours or one day. And you could go through and if you had a product manager on that and optimize every part of it, I bet you could achieve that. But the hard part isn't a person's job. It's actually all the systems and people in between it. And so I think part of the reason why I think it's been slow to get the productivity enhancement is we ship our org charts as companies naturally. That's the natural state.

1:09:47

SPEAKER_01

There's not usually a person responsible for that process. There's the legal team responsible for the contract. There's a procurement team. There's a lot of people who are involved. So I think actually we will end up reimagining our companies with the benefit of AI. [SPEAKER_00] Will we actually think of our companies as a collection of processes, have people responsible for them with KPIs, who can apply AI? [SPEAKER_00] And I think I bring it up just because that's my theory of the world. [SPEAKER_00] I might be wrong. [SPEAKER_00] I might be right. [SPEAKER_00] But I'm not sure our companies are set up to essentially absorb the benefits of AI efficiently right now.

1:10:24

SPEAKER_01

[SPEAKER_00] And we need to do that to really do so. [SPEAKER_00] But the bigger point, I think, is that there's the paradox of, well, you want more software engineers.

1:10:38

SPEAKER_00

But on top of that, most of the world isn't just digital technology.

1:10:43

SPEAKER_01

[SPEAKER_00] And so I think a lot of the people in the AGI community have only ever worked at a research lab or a software company. [SPEAKER_00] You look around and think, wow, AI is going to do all of this. [SPEAKER_00] And as they walk by the flower shop and get their coffee at the coffee shop, and you think about your local flower shop, if you took all the AI in the world and gave it to that, gave it super intelligence, how much would it impact the flower shop's operations? [SPEAKER_00] Maybe a little. [SPEAKER_00] I mean, I'm sure it would help. [SPEAKER_00] Don't get me wrong.

1:11:24

SPEAKER_01

[SPEAKER_00] But someone's still clipping the ends of the stems of the flowers, arranging the bouquets, and thanking you on your way out the door and congratulating you for your daughter's wedding or whatever it is. [SPEAKER_00] And so I think if you think about what parts of the economy can absorb intelligence really efficiently, it's certainly software. [SPEAKER_00] And we're seeing that already. [SPEAKER_00] It's finance seems particularly meaningful here because so much of finance today is just digital information. [SPEAKER_00] Everything's in digital systems now, not even just crypto.

1:11:58

SPEAKER_00

Everything's in digital ledgers everywhere. [SPEAKER_01] It still doesn't touch a wet lab. [SPEAKER_01] You still can't do a clinical trial.

1:12:15

SPEAKER_01

You still need to get a crate from this country to that country on a ship. So as a consequence, I think I'm not sure we'll see the productivity enhancement we see in software in every sector as quickly. And then on top of that, I think companies need to stop just giving co-pilot to every employee and saying we're AI now, and start to think about from first principles, what are the parts of your business that have a lot of digital workflows?

1:12:41

SPEAKER_00

[SPEAKER_01] Where can AI have a real big impact? [SPEAKER_01] And how do you actually set up your company to actually have someone accountable to drive that? [SPEAKER_01] And that feels a real big change management opportunity that most companies haven't done. [SPEAKER_01] We just push on that. [SPEAKER_01] So software engineering, I think we clearly are seeing a lot of AI productivity gains and software engineers have always loved tools and the latest tools and are diving into it. [SPEAKER_01] Then you have stuff in the flower shop where stuff that requires really good robotics that we're far away from. [SPEAKER_01] That will take a while.

1:12:58

SPEAKER_00

[SPEAKER_01] What I'm talking about is there's a big middle.

1:13:00

SPEAKER_01

And by the way, I might prefer a flower shop with the florist.

1:13:05

SPEAKER_00

[SPEAKER_01] Where can AI have a real big impact? [SPEAKER_01] And how do you actually set up your company to actually have someone accountable to drive that? [SPEAKER_01] And that feels a real big change management opportunity that most companies haven't done. [SPEAKER_01] We just push on that.

1:13:17

SPEAKER_01

So software engineering, I think we clearly are seeing a lot of AI productivity gains and software engineers have always loved tools and the latest tools and are diving into it.

1:13:20

SPEAKER_00

[SPEAKER_01] Then you have stuff in the flower shop where stuff that requires really good robotics that we're far away from.

1:13:25

SPEAKER_01

That will take a while. What I'm talking about is there's a big middle. And by the way, I might prefer a flower shop with the florist. Totally. Yeah, yeah. Just to say it. I'm not sure it solves a problem I have with my flower shop. Absolutely. I might be wrong. I might be unique in that. Yes. [SPEAKER_01] But that's my point. I think a lot of the company, a lot of the economy is actually white collar knowledge work, not coding.

1:14:03

SPEAKER_01

[SPEAKER_01] That will take a while. What I'm talking about is there's a. And by the way, I might prefer a flower shop with the florist. Totally. Yeah, yeah. Just to say it. I'm not sure it solves. I'm not sure it solves a problem I have with my flower shop. Absolutely. I might be wrong. I might be unique in that. Yes. [SPEAKER_00] [SPEAKER_01] But that's. [SPEAKER_00] [SPEAKER_01] I think a lot of the company, a lot of the economy is actually white collar knowledge work, not coding. [SPEAKER_00] [SPEAKER_01] Think of finance departments, legal departments, things like that, where you should be able to see a lot of AI uplift and a lot of AI productivity improvements.

1:15:12

SPEAKER_01

And it just feels like at the current course and speeds, we're not on track to get those productivity improvements. Well, I would, I'm not sure I'm right, but I would argue thinking about it by department rather than by processes where it's off. We can talk about the processes as well, but. Well, hear me out though on this, because if you said, I want to make the legal department more productive, so I want to make it easier to do red lines and you optimize that. [SPEAKER_00] [SPEAKER_01] But why is the contract there? What is it for?

1:15:40

SPEAKER_01

You might, if you're, for example, onboarding in a supply chain vendor and you have hundreds of them, you might actually say, actually making an abstract technology for your legal department to red line contracts more efficient is actually a harder, more general problem than for your supply chain vendors, because you might actually have very rigid rules around your supply chain.

1:15:44

SPEAKER_00

[SPEAKER_01] Let's say you're a CPG company. [SPEAKER_01] Yeah. [SPEAKER_01] And you might actually have very specific saying, look, if you want to work with us, here's our core legal terms, here's the axes of independence.

1:16:02

SPEAKER_01

And if you want to make an AI agent to automate that contract, that's actually a much more narrow problem domain that doesn't require general purpose redlining technology. In fact, if you reduce it, you could say, well, there are 10% of our suppliers where we let them negotiate their contract, but only for this spend. Let's have them go through our legal department. The rest, let's do it all with AI.

1:16:25

SPEAKER_00

[SPEAKER_01] And my point on it is, if you look at it through the lens of an end to end business process, you can turn science into engineering.

1:16:28

SPEAKER_01

And I think solving legal through AI, that's a science problem. And this is my point though, which is I think people are going through department by department. Similarly, there's not a person accountable for that end to end process. And the more you can narrow the domain that you're solving with AI, the more you can build a harness or scaffolding with existing technology to actually fully automate it. And my hypothesis is most companies just aren't set up that way. [SPEAKER_00] [SPEAKER_01] That's just not how we're organized. [SPEAKER_00] [SPEAKER_01] And as a consequence, we're all optimizing our site.

1:17:09

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] We're all just installing Copilot and Copilot's great, by the way. [SPEAKER_00] [SPEAKER_01] Didn't mean to insult it, but it's not actually. [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] And to be clear, that's the thing we're doing where, and obviously companies, good companies did this before they are a continuous process improvement. [SPEAKER_00] [SPEAKER_01] And I feel that is the best thing to do. [SPEAKER_00] [SPEAKER_01] And I think what you're saying is, there's no such thing as an AI lawyer. [SPEAKER_00] [SPEAKER_01] Instead, there's improving your commercial contracting.

1:17:35

SPEAKER_00

[SPEAKER_01] That is a thing that you can tend to. [SPEAKER_01] And even more narrowly, pick one domain of commercial contracting and solve that. And I actually think those are truly solvable. And I think the companies that really think about their business that way, I think they can see the value. And again, I'll go back to the immaturity of the applied AI market is probably one of the bigger barriers right now. And my hope is that as the applied AI market matures over the next few years, we'll see a step change in productivity. Yeah. There is a canonical way to build a Silicon Valley company. You have engineering and product and design.

1:18:20

SPEAKER_00

And you have this number of ratios of engineers to product managers and engineering managers. [SPEAKER_01] And then you have your go to market organization and you have these pipeline coverage ratios. [SPEAKER_01] And you have the product marketers and all this. [SPEAKER_01] But I find it interesting how similar so many Silicon Valley companies are to each other because they've all learned from each other, right? [SPEAKER_01] Yeah. [SPEAKER_01] [SPEAKER_00] There's a shared recipe and a shared playbook as to how to build a company. [SPEAKER_01] And it gets tweaked, but ultimately I think it's pretty good IP.

1:18:46

SPEAKER_00

[SPEAKER_01] Certainly companies are much better off with it than without. [SPEAKER_01] How is that canonical template for building a company different post AI than before? [SPEAKER_01] [SPEAKER_00] Yeah, that's a really interesting question. [SPEAKER_01] [SPEAKER_00] One is, I've always believed in the primacy of tech leads over engineering managers.

1:19:00

SPEAKER_01

[SPEAKER_00] Both Google and Facebook where I spent some of my early career both did this well where, in a product review, you weren't just talking to a manager. You were talking to the tech lead and PM, the product manager who are building the product. [SPEAKER_00] Whereas if you went to companies that produced worse software, I'd notice you sort of move up the chain of command, the military. Yes. [SPEAKER_01] I think that we will end up with individual tech leads who, because of the existence of AI agents, will become even more important. [SPEAKER_00] One is, I've always believed in the primacy of tech leads over engineering managers.

1:19:33

SPEAKER_01

[SPEAKER_00] Both Google and Facebook where I spent some of my early career both did this well where, in a product review, you weren't just talking to a manager. You were talking to the tech lead and PM, the product manager who are building the product. [SPEAKER_00] Whereas if you went to companies that produced worse software, I'd notice you move up the chain of command, the military. Yes. I think that we will end up with individual tech leads who, because of the existence of AI agents, will become even more important. And so, I think that's a really important point where, if you are a, I'll say a product engineer, I'm trying to find the right word for it.

1:20:04

SPEAKER_01

We might invent one who has taste, but didn't necessarily know CSS, who has infrastructure ability, meaning that you understand the basics of distributed systems and debugging. And you understand your customer very deeply. With the presence of Codex, you can produce amazing results. Those people are truly worth a thousand X other people because it's relatively easy to find someone who's a great infrastructure engineer. Not easy, but relatively. Yeah. Finding someone with good taste, that's relatively easy. Finding someone who also understands your customers extremely well, the nuances of the problem they're solving.

1:20:17

SPEAKER_01

Those people who can combine that will, I think, end up being able to actually produce products, capital P valuable products. [SPEAKER_01] Yes. With relative autonomy. And I wonder if it will change our view on generalists broadly. I've always identified myself as a generalist just because I've been both a software engineer in a suit, and I've gone back and forth in that world. And as companies grow, you tend towards more specialization, just because the person who's the Jack or Jill of all trades ends up not fitting in. There's not really a place for them because you're not really the deepest engineer. You're not really the best designer.

1:20:54

SPEAKER_01

You're not really a product manager. If you've been at the company for a while, we'll give you an honorary something to do. [SPEAKER_00] And you have to lead through influence and so on. [SPEAKER_00] Could that person actually endure as one of the most valuable people in these companies? [SPEAKER_00] And I think, I don't know whether it's naive optimism or true, but I actually think those people who often exist in early stage startups are often the people who get sidelined, but actually in a way that really harms the company.

1:21:33

SPEAKER_01

[SPEAKER_00] And I'm hopeful that in a world of AI agents, those generalists who, again, I think the most important part is understanding the customer need with agency, no pun intended, and empowerment can end up more powerful in the Silicon Valley company in the future. [SPEAKER_00] The exact same thing, it's right. [SPEAKER_00] The exact same thing, which is high agency, really caring about customers, just really caring generally. Yeah. High work ethic people who maybe weren't the best engineers previously or now. Those people are massively ascendant, as far as I can tell, because they suddenly got the exoskeleton. Yeah, exactly.

1:22:06

SPEAKER_01

And they always have the ideas as to what we should be doing, and this is the better way to serve the customers and everything. But now they have the way to make all their schemes real. [SPEAKER_00] I've really noticed that at Stripe. [SPEAKER_00] Well, it's interesting you talked about work ethic. [SPEAKER_00] It's addictive right now because you can do so much with the technology. [SPEAKER_00] Everyone I know who's really used it works harder because it's, wow, I could do so much. [SPEAKER_00] You think you're about to go to bed and you're, should I get an AI agent to do something? [SPEAKER_00] Am I wasting the next eight hours of my life?

1:22:42

SPEAKER_01

[SPEAKER_00] That might be a novelty that wears off, but I think it's really exciting. [SPEAKER_00] So I'm hopeful on the product engineering design side, you end up with these hyper high agency people who really deeply care. [SPEAKER_00] I really the way you said it actually. [SPEAKER_00] It's right. [SPEAKER_00] It's not just customer problems. [SPEAKER_00] [SPEAKER_01] It's care period. [SPEAKER_00] [SPEAKER_01] Yeah. [SPEAKER_00] [SPEAKER_01] Just care can end up more empowered. And I'm curious what that means for organizational structures. You know, it's—

1:23:21

SPEAKER_00

[SPEAKER_01] [SPEAKER_00] I think we have a new job role we need to invent. [SPEAKER_01] Yeah. [SPEAKER_01] It's what role do these people fit in? [SPEAKER_01] I mean— [SPEAKER_01] Hyper generalists. [SPEAKER_01] Yeah. [SPEAKER_01] Product managers, but sometimes maybe without a product, minister without a portfolio, they're just doing stuff, but now they can do much more. [SPEAKER_01] Yeah.

1:23:59

SPEAKER_01

[SPEAKER_00] [SPEAKER_01] And it's almost product designer, product manager, engineer. That's why I said product engineer, but that means something different. [SPEAKER_00] [SPEAKER_01] But it's interesting because we've talked about this. You know, you end up where the grass is always greener with org structure.

1:24:13

SPEAKER_00

[SPEAKER_01] So you go functional organization.

1:24:18

SPEAKER_01

Okay. We're going to have engineering product design. Let's go to business units.

1:24:31

SPEAKER_00

[SPEAKER_01] They're, wow, that led to silos and wars. [SPEAKER_01] One reorganization. [SPEAKER_01] Yeah.

1:24:53

SPEAKER_01

You sway back again.

1:25:00

SPEAKER_00

[SPEAKER_01] And you know, that's welcome to—

1:25:07

SPEAKER_01

Just one more reorg. Yeah, exactly. I'm a middle manager now. [SPEAKER_00] And I think that it is interesting if these people become extremely important, what does it mean to organize around them? [SPEAKER_00] Mm-hm. [SPEAKER_00] And I think it does feel something that will end up flatter just because of the amount of impact an individual can have. [SPEAKER_00] And so that feels really exciting to me, but I don't really know. It feels a blurry picture right now. [SPEAKER_00] I agree. And that's welcome to— Just one more reorg. Yeah, exactly. I'm a middle manager now.

1:26:22

SPEAKER_01

[SPEAKER_00] And I think that it is interesting if these people become extremely important, what does it mean to organize around them? Mm-hm. [SPEAKER_00] And I think it does feel like something that will end up flatter just because of the amount of impact an individual can have. And so that feels really exciting to me, but I don't really know. It feels like a blurry picture right now. [SPEAKER_00] I agree. It's very blurry. It's so interesting. You were on the Twitter board during the super interesting takeover battle with Elon Musk. What are your reflections on that experience a few years later?

1:26:59

SPEAKER_01

[SPEAKER_00] It was really interesting to be in the public spotlight. I hadn't really experienced that in my career before. I joke that no one really cares about enterprise software. I worked for Salesforce for six and a half years. Sorry, what's the joke?

1:27:05

SPEAKER_00

I worked for Salesforce for six and a half years and I don't think my mom knows what Salesforce does. And so to have something that was not just a business issue or a technology issue, but in the mainstream, I realized I didn't love that very much. [SPEAKER_01] You actually prefer enterprise software. Yeah, exactly. I'm a builder. I like to build things and have people use them. I know it sounds funny and reductive, but that's what gives me joy. So the one thing I realized is the conflict of it all. However, it turned out, victory, defeat, whatever it was, it just didn't fill my bucket very much.

1:27:27

SPEAKER_00

[SPEAKER_01] What do you make of the fact that in all these headcount debates, Elon is now running Twitter with 80, 85% fewer people? I think Nikita Beard tweeted recently that all of Eng, products, and design at Twitter is 50 people. And maybe it's really impressive. Yeah, it's been a little flaky in pockets or just at times, but mostly the service works and they have shipped new features. And I think those two statements are undeniable, but what's your takeaway from that? [SPEAKER_01] I don't know. I haven't followed as much. I didn't see that tweet as an example. Do you call it tweets still?

1:27:55

SPEAKER_01

[SPEAKER_00] Sorry. Yeah, I'm old fashioned. I hate that. So I don't know about that, but it is interesting right now because obviously a lot of that predated AI. Yeah. Yeah, it's been a little flaky in pockets or just at times, but mostly the service works and they have shipped new features. And I think those two statements are undeniable, but what's your takeaway from that? I don't know. I haven't followed as much. I didn't see that tweet as an example. Do you call it tweets still?

1:28:05

SPEAKER_01

[SPEAKER_00] Sorry. Yeah, I'm a little old fashioned. I hate that. So I don't know about that, but it is interesting right now because obviously a lot of that predated AI. Yeah. But any person who has been an individual contributor engineer knows that the size of the team does not produce linearly greater outcomes. Yes, yes. Everyone in the world has experienced that. So the idea of can you actually give individuals with good taste more agency, no pun intended, I think it's always been an enduring thing. What was Jeff Bezos, a two pizza box sort of thing?

1:28:13

SPEAKER_01

[SPEAKER_00] Sure. But then do large tech companies underrate this phenomenon? Do they pay lip service to small empowered teams and two pizza teams when maybe they should be doing more?

1:28:18

[SPEAKER_01] I think their companies largely act somewhat rationally. I can't remember who the CEO was, but it might have been the Ripley CEO just talking about the idea of being lean and agile and then there's wanting to capture market share and grow your product and grow your platform. And at the end of the day, you can be clever, but not smart. You might be so clever to think, I'm not going to have anything more than two people on these features. And if you have a competitor who maybe does something a little less elegantly, but wins, who cares that you are clever with your two pizza box team or two person team or one AI agent team or whatever it is. When someone said we're going to have a X billion dollar company with one person, I think that might have been right, but it's not.

1:28:21

SPEAKER_00

You could have had a $10 billion company if you'd hired a bit more.

1:28:23

SPEAKER_00

[SPEAKER_01] That's right. And I would actually argue the more specific thing is if all of a sudden for some clever reason, you want to prove you can, the idea that a competitor might have 10 people and beat you is probably more likely than even having a $10 billion company. And so at the end of the day, when you're building a business, especially one that's in hyper growth, which successful businesses in tech tend to be, if you are too clever and austere and going back to your point about Silicon Valley cultures all being the same, there are examples of companies that really innovated in culture. You wouldn't think of it this way, but HP, the kind of traditional open office floor plan, came from them. Facebook—

1:28:26

SPEAKER_00

Lows worked at HP. [SPEAKER_01] Oh, I didn't know that. That's interesting. And then Google offered free food to their employees, which a lot of people did. And then Facebook, a lot of them, both the layouts of offices all looked like Facebook for a long time. But then you have other companies working to innovate in HR and they spend all this time and energy on it. And in fact, the smart thing to do is just be like, we're not, it's not what we do. Let's just do the same old thing as everyone else because everything is just push button. I don't need to worry about it. And so I think it's the right question to ask for every technology company. [SPEAKER_01] Yeah.

1:28:50

SPEAKER_00

After being on the Twitter board during the Elon takeover, you were then on the OpenAI board when Sam got fired. Have you considered that you are the problem? You are bringing the drama. [SPEAKER_01] I came in after the drama there. Oh, right. You joined after. [SPEAKER_01] Oh, sorry. I was brought in as the mediator. I see. Yeah.

1:29:06

SPEAKER_01

Okay. Your hands are clean. [SPEAKER_00] I'm aligning my reputation here. Yeah. I wasn't actually on the other side of it, but I got a phone call, was it Saturday or Friday after? I came in after the drama there. Oh, right. You joined after.

1:29:14

SPEAKER_00

[SPEAKER_01] Oh, sorry. I was brought in as the mediator.

1:29:16

SPEAKER_01

[SPEAKER_00] I see. [SPEAKER_00] Yeah. [SPEAKER_00] Okay. Your hands are clean. [SPEAKER_00] I'm aligning my reputation here. [SPEAKER_00] Yeah. I wasn't actually on the other side of it, but I got a phone call, was it Saturday or Friday after? Oh, yeah. My understanding was I was the person that both the existing board and Sam agreed upon to help mediate the situation. [SPEAKER_00] What have you learned on the OpenAI board? A lot. Certainly the most interesting part is the AI research. I've never been affiliated with a true research lab before and that's fascinating to me. [SPEAKER_00] Yeah.

1:29:34

SPEAKER_01

It is very inspiring. It's very easy to grow cynical, but you can look at OpenAI, Google, Anthropic and say whose model scores better on this leaderboard. To actually go in and see this company where every single researcher is trying to make safe AGI and not come out of those board meetings as inspired is impossible. It's amazing. The other thing is it's the first non-profit board I've been affiliated with. And that's really interesting as well. [SPEAKER_00] It's a different thing, yeah. Well, I mentioned the fiduciary duty is you have a duty to the mission. [SPEAKER_00] Yeah.

1:29:41

SPEAKER_01

And that is really clarifying and interesting as well because when you're making decisions and you realize your sole duty is to ensure that artificial general intelligence benefits humanity. That's really different. It's really interesting. I've never had a fiduciary duty to a mission before. [SPEAKER_00] Yeah.

1:29:43

SPEAKER_01

So that's really interesting to me because I take those duties really seriously and reflecting in a board meeting and you're making a decision, you think about it very differently through that context. And then the other thing was because I was brought in after that crisis, there was three people on the board on the other side of that when I agreed to temporarily be the chairman. It's still there. Funny how that works. We had to grow the board essentially from scratch. [SPEAKER_00] Yeah, yeah, yeah. And so that was really interesting too, just to think about. Normally you're at one board member at a time.

1:29:47

SPEAKER_01

[SPEAKER_00] Yeah, yeah, yeah. This one was like, do you have a bulk rate? We're gonna build a board, put together a team. So you really think about. I spent time with the other two board members just really thinking about what is the composition for an OpenAI board look like? How do you represent the not-for-profit part of it? How do you represent safety? [SPEAKER_00] Yeah. How do you represent the economic impact of AI? We're doing lots of infrastructure investments. How do we find someone with that specific type of financial expertise? And so that was really rewarding as well. Just building a board, not from scratch, but effectively from scratch. Last question.

1:29:52

SPEAKER_01

[SPEAKER_00] What are your AI predictions for 2026? I think we will have some scientific breakthroughs with AI that positively break through into the mainstream press and awareness. We've already had some interesting math proofs. But I joked with one of my friends, like, until I can understand what the title means, I'm not sure it's gonna make— Yeah, exactly. And you know, it won't quite be like the Apollo landing, but I remember the Kasparov chess match and I certainly things like AlphaGo were really meaningful.

1:29:54

SPEAKER_00

Yeah. I, given the progress in math, I'm hopeful we have at least one moment of discovery that is inspiring because I think a lot of the dialogue around AI right now is economic opportunities, but also what could go wrong. And I actually think one of the main things that can go right is actually discovery in science that actually can improve the human condition. So I'm really excited for it because I think it will contextualize why so many of us are excited about this technology in a way that captures attention. So as you said, something beyond n-dimensional manifold, blah, blah, blah. And I feel not confident in that, but it certainly feels like the ingredients are there for that. I think we'll continue to see mainstream adoption of AI by both consumers and companies. That doesn't really feel like a prediction, but I think this will be really a year of adoption of agents. And we're certainly seeing that in CIRA's customer base, but I think we're going to see it more writ large. And then you already see in ChatGPT growth really unprecedented levels and things like OpenClaw. You can see that translate over to agents and the more long running and autonomous tasks. So it does feel like by the time we exit this year, can that go from a niche community to something more mainstream? It feels probable to me. And then the other thing is I think most companies in Silicon Valley won't write code by hand. And that might seem almost. It's funny that it feels obvious right now.

1:30:02

SPEAKER_00

[SPEAKER_01] Yeah. Of course. You're just nodding. Of course. Yeah, why not? But if I had said that four months ago, that would have been a bold prediction. [SPEAKER_01] Yeah. But I think that's really interesting just because that's such a fundamental state change. [SPEAKER_01] Yeah. And I say in Silicon Valley because I do think it takes a while for these tools to diffuse throughout society. Silicon Valley is insular enough that I think it will here, but I'm not sure it will happen through every company in the world yet.

1:30:20

SPEAKER_00

[SPEAKER_01] So the year of agents across businesses and just people finally getting their OpenClaw style agents and then all code written by AI as well. Yeah. It's good celebrations. Right. Thank you.

1:30:21

SPEAKER_01

Thanks for having me. [Separate interview] [SPEAKER_00] I think Nikita Beard tweeted recently that all of Eng products and design at Twitter is 50 people. And you know, maybe it's really impressive. [SPEAKER_00] Yeah, it's been a little flaky in pockets or just at times, but mostly the service works and they have shipped new features. [SPEAKER_00] And I think those two statements are undeniable, but just what's your takeaway from that? I don't know. I haven't followed as much. I didn't see that tweet as an example. [SPEAKER_00] Do you call it tweets still? [SPEAKER_00] Sorry. Yeah, I'm a little old-fashioned. I hate that.

1:30:56

SPEAKER_01

So I don't know about that, but I mean, it is interesting right now because obviously a lot of that predated AI.

1:30:59

SPEAKER_00

Yeah.

1:31:00

SPEAKER_01

But I mean, any person has been an individual contributor engineer knows that the size of the team does not produce linearly greater outcomes. [SPEAKER_00] Yes, yes. Everyone in the world has experienced that. I don't know. I haven't followed as much as the tweet as an example. [SPEAKER_00] Do you call it tweets still?

1:31:19

SPEAKER_00

Sorry. Yeah, I'm a little old fashioned. [SPEAKER_01] I hate that. [SPEAKER_01] So I don't know about that, but it is interesting right now because obviously a lot of that predated AI. [SPEAKER_01] Yeah.

1:31:38

SPEAKER_01

But any person has been an individual contributor engineer knows that the size of the team does not produce linearly greater outcomes. Yes, yes. [SPEAKER_01] Everyone in the world has experienced that. So I think the idea of can you actually give individuals with good taste more agency, no pun intended, I think it's always been an enduring thing. What was Jeff Bezos, a two pizza box thing? Sure. [SPEAKER_00] But then do large tech companies underrate this phenomenon? [SPEAKER_00] Do they pay lip service to small empowered teams and two pizza teams and extras that maybe they should be doing more? [SPEAKER_00] I think their companies largely act somewhat rationally.

1:32:11

SPEAKER_01

[SPEAKER_00] I can't remember who the CEO was, but it might have been the Ripley CEO just talking about, there's this idea of being lean and agile and then there's wanting to capture market share and grow your product and grow your platform.

1:32:16

SPEAKER_00

[SPEAKER_01] And at the end of the day, you can be clever, but not smart. [SPEAKER_01] You might be so clever to think, I'm not going to have anything more than two people on these features. [SPEAKER_01] And if you have a competitor who maybe does something a little less elegantly, but wins, who cares that you are clever with your two pizza box team or two person team or one AI agent team or whatever it is.

1:32:19

SPEAKER_01

When someone said we're going to have a X billion dollar company with one person, I think that might've been right, but it's not. [SPEAKER_00] You could have had a $10 billion company if you'd hired a bit more. [SPEAKER_00] That's right. And I would actually argue the more specific thing is if all of a sudden for some clever reason you want to prove you can, the idea that a competitor might have 10 people and beat you is probably more likely than even having a $10 billion company.

1:32:29

SPEAKER_01

And so at the end of the day, when you're building a business, especially one that's in hyper growth, which successful businesses in tech tend to be, if you are too clever and austere and going back to your point about Silicon Valley cultures all being the same, there are examples of companies that really innovated in culture. [SPEAKER_00] You wouldn't think of it this way, but HP, the traditional open office floor plan came from them. Facebook- [SPEAKER_00] Laws worked at HP. [SPEAKER_00] Oh, I didn't know that. [SPEAKER_00] That's interesting. [SPEAKER_00] And then Google offered free food to their employees, which a lot of people did.

1:32:52

SPEAKER_01

[SPEAKER_00] And then Facebook, a lot of them both liked the layouts of offices all looked like Facebook for a long time. But then you have other companies working to innovate in HR and they spend all this time and energy on it. [SPEAKER_00] And in fact, the smart thing to do is just be like, we're not, it's not what we do.

1:32:56

SPEAKER_00

Let's just do the same old thing as everyone else because everything is push button. I don't need to worry about it. [SPEAKER_01] And so I do think it's the right question to ask for every technology company. [SPEAKER_01] Yeah. [SPEAKER_01] After being on the Twitter board during the Elon takeover, you were then on the OpenAI board when Sam got fired.

1:33:17

SPEAKER_01

Have you considered that you are the problem? You are bringing the drama. I came in after the drama there. Oh, right.

1:33:36

SPEAKER_00

[SPEAKER_01] You joined after.

1:33:37

SPEAKER_01

Oh, sorry. I was brought in as the mediator. [SPEAKER_00] I see. [SPEAKER_00] Yeah. [SPEAKER_00] Post the, okay.

1:33:57

SPEAKER_00

Yeah. [SPEAKER_01] Okay.

1:34:00

SPEAKER_01

Your hands are clean. I'm aligning my reputation here. Yeah. I wasn't actually on the other side of it, but I got a phone call, was it Saturday or Friday after? Oh, yeah.

1:34:26

SPEAKER_00

[SPEAKER_01] My understanding was I was the person that both the existing board and Sam agreed upon to help mediate the situation. [SPEAKER_01] What have you learned on the OpenAI board? [SPEAKER_01] A lot. [SPEAKER_01] Certainly the most interesting part is the AI research. I've never been affiliated with a true research lab before and that's fascinating to me. [SPEAKER_01] Yeah. [SPEAKER_01] It is very inspiring.

1:34:49

SPEAKER_00

[SPEAKER_01] It's very easy to grow, not cynical, but you can look at OpenAI, Google Anthropic and say, whose model scores better on this leaderboard, but to actually go in and see this company where every single researcher is trying to make safe AGI and not come out of those board meetings as far as impossible. [SPEAKER_01] It's amazing.

1:34:58

SPEAKER_01

The other thing is it's the first not-for-profit board I've been affiliated with. And that's really interesting as well. It's a different thing, yeah.

1:35:05

SPEAKER_00

Well, I mentioned the fiduciary duty is you have a duty to the mission. [SPEAKER_01] Yeah. [SPEAKER_01] And that is really clarifying and interesting as well because when you're making decisions and you realize your sole duty is to ensure that artificial general intelligence benefits humanity.

1:35:14

SPEAKER_01

That's really different. It's really interesting. I've never had a fiduciary duty to a mission before. Yeah. So that's really interesting to me because I take those duties really seriously and reflecting in a board meeting when you're making a decision, you think about it very differently through that context. And then the other thing was because I was brought in after that crisis, there were three people on the board on the other side of that when I agreed to temporarily be the chairman. [SPEAKER_00] It's still, still there. [SPEAKER_00] Funny how that works. [SPEAKER_00] We had to grow the board essentially from scratch.

1:35:38

SPEAKER_01

I've never had a fiduciary duty to a mission before. Yeah.

1:35:41

SPEAKER_00

[SPEAKER_01] So that's really interesting to me because I take those duties really seriously and reflecting in a board meeting and you're making a decision, you think about it very differently through that context. [SPEAKER_01] And then the other thing was because I was brought in after that crisis, there were three people on the board on the other side of that when I agreed to temporarily be the chairman. It's still there. Funny how that works.

1:35:42

SPEAKER_01

[SPEAKER_00] We had to grow the board essentially from scratch. [SPEAKER_00] Yeah. [SPEAKER_00] And so that was really interesting too, just to think about, normally you're adding one board member at a time. Yeah, yeah, yeah. This one was like, do you have a bulk rate? [SPEAKER_00] We're gonna build a board, put together a team. [SPEAKER_00] So you really think about, I spent time with the other two board members just really thinking about what is the composition for an OpenAI board look like? [SPEAKER_00] How do you represent the not-for-profit part of it? [SPEAKER_00] How do you represent safety? Yeah. How do you represent the economic impact of AI?

1:36:08

SPEAKER_01

We're doing lots of infrastructure investments.

1:36:11

SPEAKER_00

[SPEAKER_01] How do we find someone with that specific type of financial expertise? [SPEAKER_01] And so that was really rewarding as well.

1:36:17

SPEAKER_01

Building a board, not from scratch, but effectively from scratch. [SPEAKER_00] Last question. What are your AI predictions for 2026? I think we will have some scientific breakthroughs with AI that positively break through into the mainstream press and awareness. We've already had some interesting math proofs. But I joked with one of my friends that until I can understand what the title means, I'm not sure it's going to make it. Yeah, exactly. And it won't quite be like the Apollo landing, but I remember the Kasparov chess match, and I certainly know things like AlphaGo were really meaningful. Yeah.

1:36:53

SPEAKER_01

Given the progress in math, I'm hopeful we have at least one moment of discovery that is inspiring, because I think a lot of the dialogue around AI right now is economic opportunities, but also what could go wrong.

1:36:53

SPEAKER_00

And I actually think one of the main things that can go right is discovery in science that actually can improve the human condition.

1:36:57

SPEAKER_01

So I'm really excited for it because I think it will contextualize why so many of us are excited about this technology in a way that captures attention. So something beyond n-dimensional manifold blah, blah, blah. And I feel not confident in that, but it certainly feels like the ingredients are there for that. I think we'll continue to see mainstream adoption of AI by both consumers and companies. That doesn't really feel like a prediction, but I think this will be a year of adoption of agents. And we're certainly seeing that in Cira's customer base, but I think we're going to see it more broadly.

1:37:14

SPEAKER_01

And then you already see in ChatGPT growth really unprecedented levels and things like OpenAI o1. You can sort of see that translate over to agents and more long-running and autonomous tasks. So it feels like by the time we exit this year, can that go from a niche community to something more mainstream? It feels probable to me. And then the other thing is I think most companies in Silicon Valley won't write code by hand. And that might seem almost funny that it feels obvious right now.

1:37:32

SPEAKER_00

[SPEAKER_01] Yeah. [SPEAKER_01] Like, of course. You're just nodding. Yeah, of course. Why not? [SPEAKER_01] But if I had said that four months ago, that would've been a bold prediction. [SPEAKER_01] Yeah. [SPEAKER_01] But I think that's really interesting just because that's such a fundamental state change. [SPEAKER_01] Yeah. [SPEAKER_01] And I say in Silicon Valley because I do think it takes a while for these tools to diffuse through society. [SPEAKER_01] Silicon Valley is insular enough that I think it will happen here, but I'm not sure it will happen through every company in the world yet.

1:37:48

SPEAKER_01

[SPEAKER_00] So the year of agents across businesses and people finally getting their kind of OpenAI o1-style agents and then all code written by AI as well. Yeah.

1:37:50

SPEAKER_00

[SPEAKER_01] Good celebrations. [SPEAKER_01] Right. [SPEAKER_01] Thank you. Thanks for having me. You know, how do you represent the not-for-profit part of it? How do you represent safety?

1:38:06

SPEAKER_01

Yeah. Um, how do you represent, you know, the economic impact of AI? Oh, we're doing lots of infrastructure investments. Like, how do we find someone with like, that specific type of financial expertise? And so that was really rewarding as well. Um, just sort of building a board, not from scratch, but, you know, effectively from scratch.

1:38:24

SPEAKER_00

Last question.

1:38:24

SPEAKER_01

What are your AI predictions for 2026? I think we will have some scientific breakthroughs with AI that positively break through into the mainstream, uh, press and awareness. We've already had some interesting math proofs. Um, but I joked with my, one of my friends, like, until I can understand what the title means, I'm not sure it's gonna make like- Yeah, exactly. Um, and, uh, you know, it won't quite be like the, you know, Apollo landing, but, you know, I remember, you know, uh, the Kasparov, you know, chess match and, and I certainly, you know, things like AlphaGo were really meaningful. Yeah.

1:39:10

SPEAKER_01

I, given the progress in math, I'm hopeful we have at least one moment of discovery, uh, that is inspiring, uh, because I think a lot of the dialogue around AI right now is economic

1:39:23

SPEAKER_00

opportunities, but also what could go wrong.

1:39:25

SPEAKER_01

And I actually think one of the main things that can go right is actually discovery in science that actually can improve the human condition. So I'm really excited for it because I think it will, uh, contextualize why so many of us are excited about this technology in a way that sort of captures attention. So as you said, something beyond n-dimensional manifold, blah, blah, blah. Um, and I feel not confident in that, but it certainly feels like the ingredients, um, are there for that. Um, I think we'll continue to see mainstream adoption of AI by both consumers and companies. That doesn't really feel like a prediction, but I think, I think this will be really a

1:40:03

SPEAKER_01

year of adoption of agents. Uh, and, uh, we're certainly seeing that in CIRA's, uh, customer base, but I think we're going to see it, uh, more writ large. And then you already see in chat TPT growth, you know, uh, really unprecedented levels and things like open claw. You can sort of see that kind of translate over to agents and sort of the more like long running and autonomous tasks. So it does feel like by the time we exit this year, can that go from a niche community to something more mainstream? It feels probable to me. And then, um, the other thing is I think most companies in Silicon Valley won't write code by hand.

1:40:43

SPEAKER_01

Um, and that might seem almost, it's sort of funny that it sort of feels obvious right now. Yeah. Like, oh yeah, of course. Like you're just nodding. Like, yeah, of course. Yeah, why not? But if I had said that like four months ago, that would've been a bold prediction. Yeah. But I think that's really interesting just because that's a such a fundamental state change. Yeah. And I say in Silicon Valley because I do think it takes a while for these tools to, uh, sort of diffuse your society. Silicon Valley is insular enough that I think it will here, uh, but I'm not sure it will happen, uh, through every company in the world yet.

1:41:15

SPEAKER_00

So the year of agents across businesses and just people finally getting their kind of claw style agents and then yeah, all code written by AI as well.

1:41:23

SPEAKER_01

Yeah. It's good celebrations. Right. Thank you.

1:41:25

SPEAKER_00

Thanks for having me.

Reading tools

Type to find a passage

Appearance
Ask this transcript

Notebook

Export Markdown

Add a note