Collection

Library

Browse your videos and channels

Video Assets

916

Tracked Channels

30

Videos 916 Channels 30

AI Engineer

AI Engineer Melbourne 2026 Keynote Livestream | Day 2

25 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: AI tools risk eroding craft, autonomy, and mastery in software engineering unless engineers deliberately choose augmentation over replacement—Jeremy Howard demos this …

Unread

AI Engineer

SWE-rebench: Lessons from Evaluating Coding Agents — Ibragim Badertdinov, Nebius

11 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Real-world coding agent evaluation requires rigorous decontamination, minimalist agent design, strong infrastructure, and constant vigilance against model cheating—not…

Unread

AI Engineer

The Art & Science of Benchmarking Agents — Vincent Chen, Snorkel AI

17 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Effective agent benchmarks require rigorous scientific foundations (task quality, distributional control, model headroom, robust methodology) plus strategic art (clear…

Unread

20VC with Harry Stebbings

Token Budgeting Panic Hits Corporate America | Cognition Raises $1BN at $26BN Valuation

88 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Corporate America is experiencing a token budgeting panic as Q1 AI spending blew through annual budgets; meanwhile, Anthropic's IPO filing, Cognition's $26B valuation,…

Unread

Stripe

10 Years of Stripe France: How enterprises pivot in the age of AI

40 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Three major French enterprises—Le Monde, Accor, and Génération Do It Yourself—demonstrate that successful AI-era transformation requires speed, fearlessness, and inter…

Unread

My First Million

The one calculation Elon runs everything on (it made him billions)

56 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Real differentiation, counterintuitive moves, and first-principles thinking (the 'idiot index' Elon runs everything on) can turn commodity industries into multi-hundre…

Unread

Every

The SaaS Apocalypse Is a Goldmine With Figma’s Matt Colyer

31 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: The 'SaaSpocalypse' narrative is backwards—AI is creating a goldmine for SaaS companies by exploding the number of software builders from ~30M developers to potentiall…

Unread

AI Engineer

How to talk to statues — Joe Reeve, ElevenLabs

28 min read

## 30-second take Joe Reeve from ElevenLabs built a viral statue-talking app in two hours using cursor, OpenAI deep research, and ElevenLabs' voice design + conversational agents APIs. The app takes a photo of a statue, …

Unread

AI Engineer

20 days of compute vs 7 hours: rethinking what state-of-the-art means — Bertrand Charpentier, Pruna

14 min read

## 30-second take Bertrand Charpentier from Pruna argues that "state-of-the-art" AI model selection is broken because people either naively trust inconsistent public leaderboards or run biased internal benchmarks—both of…

Unread

AI Engineer

Can LLMs generate Enterprise Quality Code? — Prasenjit Sarkar, Sonar

11 min read

## 30-second take Sonar's Prasenjit Sarkar argues that while LLMs achieve 80%+ pass rates on functional tests (SWE-bench), they produce enterprise-unready code with severe maintainability, security, and verbosity issues.…

Unread

AI Engineer

What if the network was the sandbox? — Remy Guercio, Tailscale

20 min read

## 30-second take Tailscale's Remy Guercio proposes using **the network layer as an AI agent sandbox**, moving auth/authz from inside containers (where agents can access their own credentials) to the network itself via …

Unread

20VC with Harry Stebbings

Mercor CEO on Why Application Layer Companies Have No Defensibility & Token Spend Exceeds Salaries

57 min read

## 30-second take Brandon Foodie, 20-something CEO of Mercor (a data provider valued at $10B+ doing over $1B in revenue), argues that the AI application layer has no defensibility because foundation models will subsume t…

Unread

Stripe

From Africa to the Andes: Building businesses without borders

3 min read

## 30-second take This is a Stripe promotional video featuring founders from Africa and South America explaining how stablecoins via Stripe solve cross-border payment friction. The actual substance is thin—three founders…

Unread

My First Million

7 bizarrely good startups that the internet has not caught up to yet

42 min read

## 30-second take This is a podcast that evaluates seven unconventional startups across health-tech, gaming, wellness, and anti-smartphone products—determining if they're "good crazy" or "bad crazy." The hosts assess Pet…

Unread

AI Engineer

How Lovable self-improves every hour — Benjamin Verbeek, Lovable

15 min read

## 30-second take Lovable (a no-code/vibe-coding platform building 200K projects/day) is attacking the continuous learning problem with two feedback loops: (1) **Lovable Stack Overflow** detects when users get stuck (sa…

Unread

AI Engineer

Task Fidelity Scaling Laws — Kobie Crawdord, Snorkel

15 min read

## 30-second take Snorkel's research demonstrates that **task quality in agentic RL training produces a 5x performance differential** (6% improvement vs. 1%) when training with high-quality versus low-quality terminal-be…

Unread

AI Engineer

What Lies Beneath the API — Benjamin Cowen, Modal

7 min read

## 30-second take Modal's Ben Cowen argues that AI startups inevitably cross from generic API use to fine-tuning as they scale and specialize—when frontier models cost more than customers pay, when latency/throughput hit…

Unread

AI Engineer

How I deleted 95% of my agent skills and got better results — Nick Nisi, WorkOS

15 min read

## 30-second take Nick Nisi (WorkOS DX engineer) built two AI agent systems—one internal (Case) and one customer-facing (WorkOS CLI)—and discovered that deleting 95% of his agent skills/prompts dramatically improved per…

Unread

AI Engineer

Spec-Driven Testing for Agents With A Brain the Size of A Planet — Steven Willmott, SafeIntelligence

11 min read

## 30-second take SafeIntelligence CEO argues that "smarter ≠ safer" for deployed agents and proposes spec-driven testing that goes far beyond eval datasets. His thesis: you need formal specifications capturing rules, on…

Unread

AI Engineer

Engineering voice agents: Latency, quality, and scale — Rishabh Bhargava, Together AI

19 min read

## 30-second take Rishabh Bhargava (Together AI, voice team lead) delivers a practical engineer's guide to production voice agents. His thesis: building quality real-time voice is no longer a research problem—it's an **…

Unread