AI Engineer

Your Agent Is an Infinite Canvas — RL Nabors, Dressed for Space

1130 summary words 5 min summary Watch video

Start with the signal

5 min read

Summary

30-second take

Rachel Lee Nabors (standards veteran from Mozilla/W3C/Microsoft, now at Arise) argues that the browser is an infinite canvas for agentic experiences, not a dying relic. Her thesis: MCP servers can be hosted via HTTP like normal websites (not just local STDIO), MCP apps turn agents into interactive UI platforms (not just chat), and WebMCP will let browser-based agents call JavaScript functions directly from pages they visit. She demos a webcomic site retrofitted with MCP tools, an embedded comic reader app inside Claude, and WebMCP-powered page navigation—showing how CSS/JS become the language of agent UX, not just websites. The talk is opinionated, practical, and pushes back on "chat as the future" orthodoxy.

Key takes

  • HTTP-based MCP servers are easier to distribute than STDIO: Users paste a URL into Claude settings instead of wrangling JSON config files with command-line args. Works with serverless/edge functions (Vercel, Cloudflare). Nabors hosts her comic archive's MCP tools at /mcp endpoint; users add it like a normal API.
  • MCP apps embed interactive UIs inside agents: Single HTML files (with Base64-encoded assets) that render as iframes in Claude. She built a comic reader that lets users navigate pages, toggle transcripts, show comments—all inside the chat interface. Apps can call MCP tools but can't access localStorage or external networks without CORS. This challenges the "chat is the CLI of agents" assumption.
  • MCP resources are spec'd but invisible in UIs: Nabors wants to pre-load comic transcripts as context for meta-analysis, but no agent harness exposes resources to users. She calls out the gap: agents won't use tools to fetch markdown docs (it burns tokens), so resources are the right vehicle—but clients haven't implemented them.
  • WebMCP turns browser pages into MCP-like tool servers: Two flavors: declarative (add tool-name attribute to forms) or imperative (navigator.modelcontext.registertools). Agents in browsers can call JS functions directly instead of taking screenshots or parsing DOM XML. Example: a "next page" function the agent invokes by voice command. Still early/not standardized, but originated at Amazon to solve auth issues.
  • Design systems and bundling matter for MCP apps: Share fonts/CSS from the same server, set CORS policies, use tools like Vite single-file bundler. Links require host permission ("mother may I"), external resources need CSP config or they're blocked. Apps can call tools but should mark visibility to prevent models from trying to render raw JSON.

Useful details

  • Nabors' background: 400k weekly readers for webcomics (2006 iVillage era), W3C Web Animations API, Mozilla Firefox DevTools, Microsoft Edge PM, React docs (react.dev/reactnative.dev). Now principal DX engineer at Arise (evals platform). Newsletter: The Agentic Web.
  • Demo tools on comic site: list_comics, list_storylines, list_characters, search_comics (searches transcripts), search_by_character, get_transcript (returns markdown). All return JSON except transcript. Tools mirror website navigation.
  • MCP app workflow: Define tool with meta: { ui: "https://..." } attribute. Returns single HTML file with everything embedded. Sandboxed iframe, no localStorage, async tool calling only. Example: comic reader with forward/back navigation via tool calls, toggling transcript/commentary.
  • WebMCP imperative example: navigator.modelcontext.registertools({ name: "next_page", description: "...", inputSchema: {...}, execute: () => { / grab link[rel=next] and navigate / } }). Demoed with MCP-B browser extension (debugging tool that shows exposed tools, mimics future Gemini/in-browser agents).
  • Web APIs she highlights for agent UX: Web Speech (text-to-speech demo in comic reader, "sounds terrible" but zero dependency), Animation, Audio, Canvas, WASM, CSS. Already standardized and waiting.
  • Security/privacy: Explicitly out of scope for this talk. CORS and CSP are footguns; she warns to check CSP if apps render blank.
  • Repo note: She glazed her comic images to prevent training use; repo has details on content protection.

Caveats / counterpoints

  • WebMCP is not MCP: "WebMCP is to MCP as JavaScript is to Java"—inspired by hype but not spec-compliant. Standards may diverge. It's "just tools for agents in your browser," not a full MCP implementation.
  • No client implements MCP resources visibly: She openly begs harness developers to expose resources in UIs. Without this, the "pre-prime context with docs" use case is broken. Agents refuse to call tools just to add markdown, so resources are essential but missing.
  • MCP apps are islands: Single HTML file, no dynamic external fetches unless CORS is set. No localStorage, no network access without server permission. This limits interactivity compared to normal web apps.
  • Chat UI skepticism is strong but not universal: Nabors calls chat "the lowest common denominator" and compares it to DOS CLI (her COBOL-programmer mother forced her to type cd blah.exe instead of double-clicking). But she admits chat works well "if you already know what the system can do" (e.g., Linear for ticketing). Her pushback is about onboarding and discovery, not dismissing chat entirely.
  • HTTP MCP adoption is low: She asks the room who hosts MCP servers on the web—almost no hands. Most people use local STDIO. HTTP is easier for users but requires hosting infra.
  • Transcript quality: This one is clean, substantive, and well-structured. No filler or music. Highly usable.

Ken relevance

  • Agent UX evolution: If you're building agent harnesses or MCP apps, Nabors' "chat is the CLI" framing is a forcing function. Ken's agent systems could differentiate by embedding richer UIs (MCP apps) instead of relying on text-only chat. Her design system + bundling advice is tactical for any team shipping agent UX.
  • HTTP MCP servers for distribution: If Ken is shipping MCP tools for customers or partners, HTTP hosting (not STDIO) simplifies onboarding. Just paste a URL. This maps to SaaS distribution models Ken knows from Carta/RocketSpace. Serverless/edge deployment (Vercel, Cloudflare) is a natural fit.
  • WebMCP for browser-agent integration: If Ken's agents need to scrape or navigate websites (e.g., for content ingestion or ops workflows), WebMCP could replace screenshot-based navigation or DOM parsing. Early but high leverage if adopted. Watch the standards group and MCP-B extension.
  • Resources gap is a blocker: If Ken wants agents to ingest large doc sets (e.g., internal wikis, product docs) without burning tokens on tool calls, he'll hit the same wall Nabors did. Arise (her employer) focuses on evals; this could inform how you test context management in agent systems.
  • Content protection angle: Her glazing/anti-training setup (linked repo) is relevant if Ken's content IP or client data is at risk of scraping for model training. Adjacent to agent security concerns.

Watch verdict

Watch fully. Nabors is a rare standards veteran with agent product intuition—she bridges web primitives and agent UX in ways most talks miss. The HTTP vs STDIO comparison, MCP app footguns, and WebMCP demo are immediately actionable if you're shipping agent tools or integrations. Even if you don't care about comics, the "infinite canvas" framing challenges chat-only orthodoxy in a way that will shape Ken's thinking on agent UX and distribution.

Search this video Resume transcript at 0%
Full transcript 4043 words · 20 min read
0:15

SPEAKER_01

Even though I know some people are still coming in, I've also been informed to start directly on time. So much apologies to anyone who is hustling up here from one of the talks downstairs. You should have known this was the best track and hung out here from hour zero. But by all means, come in, have fun. Hi there. I am Rachel Lee, Rachel Lee Neighbors. Neighbors like nearest neighbors, but not spelled that way. I have worked on the standards that powers today's web with Mozilla on Firefox DevTools, W3C on web standards like the Web Animations API. I've even been a PM on Microsoft's Edge browser.

0:47

SPEAKER_01

And you may know me from my pandemic babies on the React team, react.dev and reactnative.dev. All right. So I've spent the last three years consulting with AI startups and LLM slash big browser companies, some of whose logos you may recognize, most of whom you will not, on all things web, AI, and UI. Recently, I've just become newly minted principal developer experience engineer at Arise. Have you ever had your CTO ruin your agentic flow with a slight change of prompt or an LLM migration? Have you ever been that CTO? Anyway, if that is you, you need to check out Arise. Evals are your friends. Check them out.

1:28

SPEAKER_01

Now, before my big tech career, I made web comics for teenage girls on a part of the iVillage network website. I don't think anyone remembers iVillage, but it was a big deal back in the day. I had 400,000 weekly readers. It was powered by ads and I learned how to build for the web so I could reach those people where they were. It was 2006 and it was a mood, let me tell you. Everyone online was edgy. We were on MySpace where we had 14 year olds talking about music with 42 year olds. And as you can see, I was really into the goth and emo scene. I was the only goth in my village.

1:59

SPEAKER_01

When I got into the web, I believed that the browser is not just a document reader, despite what the CSS wonks will try to get you to believe. It is an infinite canvas that can render anything, not just documents, but video, audio, whatever you need, there's an API for it. And I have been working to that end on the web ever since. I worked on that web animation API with the W3C, creating such insane demos as the interactive Alice in Wonderland, and generally made the web do things that it was not supposed to do. And now I write about how agents and AI are changing how we browse and consume content on my little newsletter, The Agentic Web.

2:19

SPEAKER_01

And I do a fair bit of shenanigans with agents myself. This is ASA. It stands for anti-social social agent. It's an MCP app that lets you use Twitter and Blue Sky next to each other without actually having to visit the sites or be subjected to other people's algorithms. Anyway, check it out if you like it.

2:45

SPEAKER_01

I'm here today to show you how, with the power of the browser's primitives that already exist and have already been spec'd out right now, you can turn your agent into an infinite canvas, or at least get your website into agents, one way or the other. So I have this habit every year when I'm thinking, what do I want to make talks about? What do I want to do? I look at the one thing on my laundry list that I keep putting off. In this case, my webcomics site, the archive site that I put together at 2010 and said, okay, you're good there, I gotta go now.

3:02

SPEAKER_01

And it's starting to get 404s, starting to get some broken images, maybe I shut down some clusters here and there, and it's like, oh dear, the CDN has failed. Anyway, it needed some repair, and I was thinking, well, don't just repair it, make it future proof. If everyone is working in agents these days, let's meet them where they are. I'm always meeting my audience where they are, and I still get fan mail from the women who grew up with my comics, so maybe I can get them to try out agents too. Alright, so just as a cringe warning, before you go to rachelthegreat.com, 2000s were a vibe, old comics, younger Rachel Lee, it's just a little cringe, okay?

3:24

SPEAKER_01

I want to honor the fans, and also I'm working a lot with copyrighted images here, and I don't want to get sued by Marvel, so I'm going to use my own stuff. The nice thing about having a former career as an artist is I can do whatever I want with my art, no questions asked. Of course, I did have to hose them down with glazing to make sure nobody else would. But that's another story. Check out the repo at the end of this if you want to know more about protecting your content from being used for training purposes. Now, I wanted to make my comics available and navigable. One server, but three different kinds of clients.

3:44

SPEAKER_01

Humans in browsers, humans in agents, and agents in browsers using the web on behalf of their humans. I know, it's complicated, they've got this polycule thing going on. So what we're going to cover today. MCP tools and transports, how to host them on the web, just how you would a website. MCP apps, how do they work? You've already been told a lot about this, I'm just going to show you a really cool example and point out the foot guns. And WebMCP, coming to a browser near you. Before we go, how many of you currently host an MCP server? On the web, not locally? Okay, well I love you guys. Everyone else, you're going to learn. And how many of you have made an MCP app?

4:40

SPEAKER_01

Alright, how many of you want to make one? This is going to be the next web pages. And lastly, has anyone here played with WebMCP? Okay, well we'll see if we can change that. So you're here. I assume you know what an MCP server is. I'm not going to go into them too much. So we're going to get started with transports. Now, if you've built for the web, I guess I should ask, how many of you have built websites and hosted them in your lifetimes? Alright. So you're familiar, you're probably familiar with HTTP, but you might not be familiar with the concept of transports. Turns out there's more than one protocol you can use to communicate from one surface to another.

5:28

SPEAKER_01

Transports are how MCP servers communicate with agents. Okay, well we'll see if we can change that. So, you're here. I assume you know what an MCP server is. I'm not going to go into them too much. So, we're going to get started with transports. Now, if you've built for the web, I guess I should ask, how many of you have built websites and hosted them in your lifetimes? Alright. So, you're familiar, you're probably familiar with HTTP, but you might not be familiar with the concept of transports. Turns out there's more than one protocol you can use to communicate from one surface to another. Transports are how MCP servers communicate with agents.

6:04

SPEAKER_01

And, there's an important difference between when you should use either of these. So, STDIO stands... God, it sounds terrible when you read it out. Standard Input Output, you can call it... Studio. Studio. Like the Phil Collins song, because we're in England, okay? [SPEAKER_02] Studio. We're going to call it that for the rest of this time. Unlike my German proofreader who was like, I don't think you called it Studio. I was like, dude, do you want me to call it anything else? I think we're going to have problems. Alright.

6:50

SPEAKER_02

[SPEAKER_01] So, the server, it runs as a local process.

6:51

SPEAKER_01

It's spawned by the client. This is why when you're wiring up a studio-based MCP app, it's always a string of JSON with a bunch of command line inputs that most of your users don't want to use, don't want to configure. Communication, it happens via, as I just said, standard input and output. And, the server stays alive throughout the session. HTTP is a little different because we're running over the web. We are not touching with a ten-foot pole, by the way, any of the security or privacy concerns with either of these layers. That is beyond the scope of today's conversation. But, HTTP, the server runs as a web service.

7:17

SPEAKER_01

It listens at an HTTP endpoint, which you may be familiar with from API development. And, communication happens via our best friend, HTTP POST requests. And, it works well with serverless setups. You know, Vercel and Cloudflare keep trying to sell you those edge functions. Well, we've got something to put in them now. Alright. With Studio, you have to update a config file with a command to spin up a local server. But, with HTTP, your user's experience is as simple as going over to Yon's settings. Okay, it is buried a little bit. But, trust me, when you get... Oh, my God. Please, if anyone from Anthropic is watching this, help. Give it a name. Put in a URL.

8:12

SPEAKER_01

And, it is now... Come on. Hurry. Come on. Come on. Come on. Passed Rachel Lee. But, you can see I can actually type the dang thing out. It's very... That's not how you spell it. There you go. It's very... There you go. I put it at slash MCP. You hit add. And, now, there it is. It's a connector. And, I go ahead and go in and I give it all the permissions to always allow, because I know this one. But, it's not easy. If you're using an HTTP server for your MCP tools for the user to get them into their, in this case, Claude. Alright. So, these are the tools I wanted to make available. They look a lot like what's available when they visit the website.

8:42

SPEAKER_01

They're very similar to that navigation. And, they all return structured data in the form of JSON. Except for, get transcripts here. And, we'll talk a little bit more about this guy in a little bit. But, point is, list the comics. List the storylines. Because, the comics include different story arcs and endings and beginnings. List the characters. Search said comics. Which is actually searching the transcripts. That's a different talk. And, search by character. Because, I love the cat. I want to see all the cat. And, the transcript is different. It returns markdown. Boom. Alright. Let's take a look at these babies in action. These are just the tools.

9:30

SPEAKER_01

We haven't gotten to the interactive part. Bear with me. I have a narrative arc. Mmm. Alright. It is a bit wordy. Everyone who's chats the future. I'm like, it is not the future. This is thank you for having a long discussion with me, Claude. Yeah, you did find that I make the comics. Good Claude. Good baby. [SPEAKER_01] I would like to make the full transcripts and texts available to the context for meta-analysis of themes and character arcs, etc. Instead of asking Claude to get transcript from 533 different comics. It does, it is not an efficient use of context to do that.

9:59

SPEAKER_01

I would really love to use MCP resources, which are the perfect vehicle for this and skills and so many other things. But the spec is loosely defined, it is loosely implemented, and I cannot find a client where I could actively get to these resources. They're on the server. They just ain't visible in your UI. So, please, if anyone here is listening who builds an agent harness, go, even bare bones, I'll take it. Give me something to work with here. Alright. So, I would love for one or more clients to implement resources. Agents really don't like using MCP tools to go add Markdown to their context.

10:30

SPEAKER_01

If you've ever maintained a documentation site, and you thought, I will add the MCP tools to the documentation site, that shall solve the problems with the coding agent knowing how to, no, no, rather die in a fire than go call your documentation by MCP tools. So, that is not how they were intended to be used by God. So, resources could also allow harnesses to decide when to pre-prime that context with docs. Like, you're switching over in React mode. Good. Alright. So, I would love for one or more clients to implement resources. Agents really don't like using MCP tools to go add Markdown to their context.

10:50

SPEAKER_01

If you've ever maintained a documentation site, and you thought, I will add the MCP tools to the documentation site, that shall solve the problems with the coding agent knowing how to, no, no, rather die in a fire than go call your documentation by MCP tools. So, let, that is not how they were intended to be used by God. So, resources could also allow harnesses to decide when to pre-prime that context with docs. Like, you're switching over in React mode. Good. Go get all the React resources from React's MCP server. Thank you. Goodbye. Alright. Little tirade over. That was your public service announcement. Tools. They can return plain data. JSON, text, markdown.

11:34

SPEAKER_01

They can also return an interactive app. App, app, app, app, app, app. So, I wanted to put a comic reader in your agent. So, while you're reading, you can read some things with some pictures instead of a long wall of text. It's, Claude, you are not my best friend in crisis. [SPEAKER_01] Why are you wall of texting me right now? It's been said that chat is the lowest common denominator of the user experience. That it is to the future of agentic experiences what the CLI was to software. My mother programmed in COBOL, and when at school they were making us double click on icons to access video games, she was, we do not do that in this household.

11:51

SPEAKER_01

[SPEAKER_01] We open DOS and we do CD blah blah blah dot exe. That is how programmers do things. But we all know that you do not do that on an iPhone. You tap and you point and grunt, right? So, this is probably a phase for us developers. There is a current, what I call the chat box landing page or starfish design. Because it sits there and it lets the user do all of the work. And you have to have a lot of context in your head when you land. You have to know what this thing can do. If you are on Linear, you are, I know this is a ticketing system. I am pretty sure I want to see X, Y, or Z. It is great if you are already familiar with the system.

12:36

SPEAKER_01

But otherwise, it can be challenging not to have any visual cues or guidance. It is, these text games were really popular back in the day. But I would not want to have to do my taxes by being, check mailbox. Is there mail? Whose house is this? That is not great. Asking your poor customer to do a bit too much discovery. So, fortunately, we have MCP apps. Inline, in-agent, rich media experiences created by bundling of HTML, CSS, and JavaScripts into a single file. This sounds so familiar. I am having flashbacks to DHTML. So, this is GitPage. GitPage is my MCP tool baby. And I love it very much. This is it in action. It returns an MCP app.

13:31

SPEAKER_01

I say, yo, I want to read Crow Princess the comic to the Starfish chat box of everything. And, of course, Claude's getting chatty. It's, I'll help you read. Let me just check all of my MCP things. Wow, I found it in the other comics collection. Let me go get Paige that. Here we go. Okay, thank you. And, check it out. I can show the commentary. I can show the comments. That's pretty cool. Those are available on the website. It's everything the website can do. It looks just as good as the website. You can navigate forward and backward. But the fun part is, you go up here, you can click on text mode. And, boom, it shows you the transcript.

14:10

SPEAKER_01

So, you could just read it if you wanted to. That'll come back later. So, yeah. It's a little mini website. Comic reader. But it's using all the resources from the website. Okay. Now. How do you do that? Well, I defined the tool as get Paige. Give it the description. There's an input schema. Of course, I'm going to have to take the comics ID, the storyline ID, the page number. So, it can really hone in on where to get this. To be honest, it just needs a storyline ID and the page number. I'm going to refactor that. And, of course, all of this stuff is required. But, most importantly, down here, where the magic happens, is you pass this little meta attribute. Give it UI.

14:43

SPEAKER_01

And, you point it. You point it at a universal resource locator. Boom. Right there. Love it. Okay. Important thing to keep in mind is that MCP apps are a little special. They're not websites. They're an island. They're single HTML files. So, everything that you need there has to be embedded. Base64. If you're going to link to something external, a cached CDN font file or something, you've got to add that to a CORS policy. So, this means you're not going to be dynamically pulling in stuff from other people's servers unless you really are certain about that and enabling it in your app. It's sandboxed in an iframe. You don't get local storage with this.

15:16

SPEAKER_01

So, you're not going to be tracking state across iframes. You're stuck with it. Async tool calling. Okay. And no network access. You have to ask server to do things for you with the call server tool. This gets into a real mother may I situation, which I'll cover in the gachas. So, what does this mean? It means that it becomes super useful to build your MCP app with your design system. In this case, I'm sharing fonts, CSS, and everything from the same server. It's sandboxed in an iframe. You don't get local storage with this. So you're not going to be tracking state across iframes. You're stuck with it. Async tool calling. And no network access.

15:56

SPEAKER_01

You have to ask server to do things for you with the call server tool. This gets into a real mother may I situation, which I'll cover in the gachas. So what does this mean? It means that it becomes super useful to build your MCP app with your design system. In this case, I'm sharing fonts, CSS, and everything from the same server. Got the CORS policy set up and everything. And HTML and JavaScript components can be bundled. I'm using Vite single file for this. So if you already have a design system, put your paw up if you do. Okay, good. All the rest of you, God, I'm sorry for you. But a design system is a thing of beauty. Bonus round. Links require host permission.

16:35

SPEAKER_01

Remember I said mother may I? You can't just window open or HRF. You got to be like app ref current open link pretty, pretty please. Remember I said external resources needed CORS. I got this thing called the CSP or content security policy, fonts, images, other external resources. They're blocked by default. So without that CSP config, the browser would just not load those things. So if you're getting a blank look and you're wondering what's janked up, check your CSP. And unlike MCPUI, which you may be familiar with, MCP apps can call tools. So I could do that forward and backward navigation with a tool call.

17:00

SPEAKER_01

But if you're doing that, remember some tools were not meant to be viewed as a big fat pile of JSON. You want to set a visibility app there just to make sure that the model's not going to try to use that. And return text. It's not going to go, oh, look at this lovely forward and backward. I will use that. Soccer arm. Don't do that. So people are still using browsers, right? They're smarter, more accessible browsers coming out all the time. And big and small browsers are shipping with agents built right into them. Sadly, they still struggle with navigating websites because they rely on taking screenshots and guessing how to use the navigation.

17:36

SPEAKER_01

Or traversing the DOM, which burns tokens chewing up XML, which is really not a great use. Both of these are compute intensive because you're using visual models or you're just eating a bunch of HTML to find out where the next URL is. Wouldn't it be great if they just had a way to call the same functions and links you already have in your JavaScript directly from the page that they're visiting? Well, with WebMCP, they can do just that. It makes every HTML page a mini MCP tools server. However, I just want to point out that WebMCP is to MCP as JavaScript is to Java. Inspired by riding the hype wave. But if you actually look under the hood, it's taking inspiration.

18:11

SPEAKER_01

It is not a one-to-one compliance with the MCP spec. It's really just tools for agents in your browser. So the specs might diverge. Keep that in mind. It comes in two flavors. Since my comics are not forms except for search, I use mostly the imperative model. But with declarative, you can just add tool name and tool description attributes to a form. If you're using JavaScript or functions to process that form, you basically told the browser to expose that form as a tool to the agent. Great for sites with lots of forms already. If you're one of those people whose site is a dashboard, use the imperative model. You pass navigator.modelcontext.registertools with parameters.

18:49

SPEAKER_01

It's great for API calls, workflows, data transformation. I used a lot of those. Now, if it's available, if you find model context in navigator, you can say, I'm going to register a tool. You give it a name. You give it a description. Give it an input schema. It's so similar to MCP. Amazing. How can this not be the same thing? But you also pass it an execution callback. In this case, this is a next page. This one basically grabs the link rel next from the page and passes it as the thing that you're supposed to navigate to next. So you could call this to navigate forward and backward on the page with a browser without having to click around. This is new.

19:36

SPEAKER_01

[SPEAKER_01] So I can only help you imagine how it works because the standards bodies are still fighting back and forth on it. I'm going to show you how to use it using the MCP B extension. This is a debugging extension. You can see all of the little tools are exposed over there. I got a little chat window. This imitates what people might see in Gemini or some future in-browser agent. In this case, I believe I am saying, read this to me. It returns the transcript. And then I might say, I might type in, oh yeah, thank you. I love that transcript. I might type in, go to the next page. I was speaking into the microphone at this point. That's fine.

20:05

SPEAKER_01

And when I hit go, it takes me to the next page. That's just the model calling that function on its own. [SPEAKER_01] That's the wow. Don't applaud yet. So the best place to get started to check this out is actually with this interview I did with the creator of WebMCP. It actually started at Amazon. This was something he made to get around a bunch of auth issues. And I love that I screen captured it. Window.navigator.registertool. There's standards community group. All very lovely. It's not fully supported yet, but you find all the resources in the commentary on that article. That's just the model calling that function on its own. Yay! That's the wow. Don't applaud yet.

20:47

SPEAKER_01

So, the best place to get started to check this out is actually with this interview I did with the creator of WebMCP. It actually started at Amazon. This was something he made to get around a bunch of auth issues. And I love that I screen captured it. Window.navigator.registertool. There's standards community group. All very lovely. It's not fully supported yet, but you find all the resources in the commentary on that article. So, I started my journey with the agentic web with this talk two years ago. And I was thinking the browser is going to die. But what people didn't take away from the clickbait was that the web isn't dead. It's just changing. It's evolving.

21:16

SPEAKER_01

CSS and JavaScript aren't just the language of the web. They're the language of interactive experiences on agents. All right. Here it is in action. I want to show you that there is a version of this that talks. Brenda looks skeptical while Rachel continues explaining Rachel. Ha. Gore left out some important tidbits. [SPEAKER_00] Rachel. [SPEAKER_00] What about how plant life will respond to the extra carbon dioxide? [SPEAKER_00] Panel two illustration shows tall plants and vegetation growing vigorously. [SPEAKER_00] I could keep going. [SPEAKER_00] But that's just the web speech API there. [SPEAKER_00] It sounds terrible. I'm sure you can do something great with 11 Labs.

21:59

SPEAKER_01

It's the browser's built-in text-to-speech capability.

22:01

SPEAKER_00

[SPEAKER_01] Zero dependency solution. [SPEAKER_01] No inference required for that. [SPEAKER_01] It's right there waiting for you to use it. [SPEAKER_01] So are all the other amazing APIs available on the web today. [SPEAKER_01] So all these APIs, web speech, animation, audio, canvas, WASM, CSS and more. [SPEAKER_01] They're already there waiting for you to make something with them.

22:14

SPEAKER_01

And I just can't wait to see what you do with this infinite canvas. It's the browser's built-in text-to-speech capability. Zero dependency solution. No inference required for that. It's right there waiting for you to use it. So are all the other amazing APIs available on the web today. So all these APIs, web speech, animation, audio, canvas, WASM, CSS and more. They're already there waiting for you to make something with them. And I just can't wait to see what you do with this infinite canvas. Applause.

22:49

SPEAKER_01

Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. Applause. . Applause. you

Reading tools

Type to find a passage

Appearance
Ask this transcript

Notebook

Export Markdown

Add a note