AI ENGINEER · THUMBNAIL LAB
A growing contact sheet of YouTube thumbnail experiments since Jan 1, 2026—wins, ties, and inconclusive results included.
Showing 124 of 124 scraped tests
We updated your video to use the winner
Computer-use models will agentify the web, not APIs — Dhruv Batra, Yutori
Computer-use models will agentify the web, not APIs — Dhruv Batra, Yutori
We updated your video to use the winner
Building Turbopuffer: Gergely Orosz (@pragmaticengineer ) × Simon Eskildsen (CEO)
Building Turbopuffer: Gergely Orosz (@pragmaticengineer ) × Simon Eskildsen (CEO)
We updated your video to use the winner
Bringing agents onto the world wide web — Paul Klein IV, Browserbase
Bringing agents onto the world wide web — Paul Klein IV, Browserbase
Codex and Subagents — Vaibhav Srivastav & Katia Gil Guzman, OpenAI
Full Workshop: OpenAI Codex — Vaibhav Srivastav & Katia Gil Guzman, OpenAI
OpenAI Codex Masterclass — Vaibhav Srivastav & Katia Gil Guzman
We updated your video to use the winner
The New Primitives: Building AI Native Software — Kwindla Kramer, Daily
The New Primitives: Building AI Native Software — Kwindla Kramer, Daily
Studio A/B test
How Juries and Librarians Can Solve GTM's AI Trust Problem — Alex Bauer, Upside.tech
Design Patterns for AI Trust: Juries, Libraries, and Agent Tiers — Alex Bauer, Upside.tech
Solving GTM's AI Trust Problem — Alex Bauer, Upside.tech
From RL to IRL — Gaurav Mishra, Amazon AGI Lab
From RL to IRL — Gaurav Mishra, Amazon AGI Lab
Studio A/B test
AI on Android: Ask me Anything — Florina Muntenescu & Oli Gaymond, Google DeepMind
GenAI MLKit — Florina Muntenescu & Oli Gaymond, Google DeepMind
Gemini Nano on device — Florina Muntenescu & Oli Gaymond, Google DeepMind
The test completed with no winner
Your LLM Stack Is a 2008 Database With Better Marketing — Lovina Dmello, NVIDIA
Your LLM Stack Is a 2008 Database With Better Marketing — Lovina Dmello, NVIDIA
The test completed with no winner
How Forward Deployed Engineering is done at Kepler — Vinoo Ganesh
How Forward Deployed Engineering is done at Kepler — Vinoo Ganesh
Better Agent Auth — Bereket Habtemeskel & Paola Estefania, Better Auth
Full Workshop: Better Agent Auth — Paola Estefania, Better Auth
Full Workshop: Better Auth — Paola Estefania, Better Auth
We updated your video to use the winner
How to Kill the Code Review — Ankit Jain, Aviator
How to Kill the Code Review — Ankit Jain, Aviator
We updated your video to use the winner
How Forward Deployed Engineering is done at Factory — Eno Reyes
How Forward Deployed Engineering is done at Factory — Eno Reyes
Results with very similar performance
Open Source Is Dead. Long Live Open Source. — Saoud Rizwan, Cline
Open Source Is Dead. Long Live Open Source. — Saoud Rizwan, Cline
We updated your video to use the winner
The Dark Arts of Web Automation: Teaching Agents to Use Websites Like Humans — Corey Gallon, Rexmore
The Dark Arts of Web Automation: Teaching Agents to Use Websites Like Humans — Corey Gallon, Rexmore
We updated your video to use the winner
Anthropic's CCA Exam as a Field-Guide for Agentic Engineering — Frank Coyle, UC Berkeley
Anthropic's CCA Exam as a Field-Guide for Agentic Engineering — Frank Coyle, UC Berkeley
We updated your video to use the winner
Realtime multiplayer, automation, and you! — Idan Gazit, GitHub
Realtime multiplayer, automation, and you! — Idan Gazit, GitHub
We updated your video to use the winner
Codex, Behind the Harness — Dominik Kundel, OpenAI
Codex, Behind the Harness — Dominik Kundel, OpenAI
We updated your video to use the winner
LLM Knowledge Bases: a practical guide — Ben Holmes, Warp
LLM Knowledge Bases: a practical guide — Ben Holmes, Warp
Your Agent Is an Infinite Canvas — RL Nabors, Dressed for Space
The Case Against 'Starfish' UI: Your Agent Is an Infinite Canvas — RL Nabors
Introducing WebMCP: Agents in the Browser — RL Nabors
You can't just one shot it — Mehedi Hassan, Granola
How Granola Builds — Mehedi Hassan, Granola
Feedback Loops are All You Need — Mehedi Hassan, Granola
The test completed with no winner
"I've never seen anything scarier than an LLM with tool calls." — Erik Meijer aka @HeadinTheBox
"I've never seen anything scarier than an LLM with tool calls." — Erik Meijer aka @HeadinTheBox
We updated your video to use the winner
Benchmarking Coding Agents on New vs Legacy Codebases — Denys Linkov, Wisedocs
Benchmarking Coding Agents on New vs Legacy Codebases — Denys Linkov, Wisedocs
Studio A/B test
Morgan Stanley's ALPHALAB: Multi-Agent Research Across Optimization Domains — Brendan Rappazzo
Morgan Stanley's AlphaLab: Open Source Multi-Agent Quant Research — Brendan Rappazzo
ALPHALAB: Multi-Agent Research Across Optimization Domains — Brendan Rappazzo, Morgan Stanley
We updated your video to use the winner
From fork() to Fleet: Designing an Agent Sandbox Cloud — Abhishek Bhardwaj, OpenAI
From fork() to Fleet: Designing an Agent Sandbox Cloud — Abhishek Bhardwaj, OpenAI
From fork() to Fleet: Designing an Agent Sandbox Cloud — Abhishek Bhardwaj, OpenAI
We updated your video to use the winner
DeepSWE: A Contamination-Resistant Coding Benchmark — James Shi, Datacurve
DeepSWE: A Contamination-Resistant Coding Benchmark — James Shi, Datacurve
Skills at Scale — Nick Nisi and Zack Proser, WorkOS
Full Workshop: Skills at Scale — Nick Nisi and Zack Proser, WorkOS
Full Walkthrough: Writing & Using Skills — Nick Nisi and Zack Proser
Results with very similar performance
Computer Use at the Edge of the Statistical Precipice — Pierluca D'Oro, Programma Labs
Computer Use at the Edge of the Statistical Precipice — Pierluca D'Oro, Programma Labs
Vibe Engineering Effect Apps — Michael Arnaldi, Effectful
Vibe Engineering: Code over Docs — Michael Arnaldi, CEO, Effectful
How I code Effect Apps — Michael Arnaldi, Creator, Effect
We updated your video to use the winner
Gadgets: Personal app vibe coding that is actually safe — Kenton Varda, Cloudflare
Gadgets: Personal app vibe coding that is actually safe — Kenton Varda, Cloudflare
Gadgets: Personal app vibe coding that is actually safe — Kenton Varda, Cloudflare
Studio A/B test
Bounded Autonomy: Between Free Will and Determinism — Angus J. McLean, Oliver
How our agents create 4,000 assets a day for 200+ brands — Angus J. McLean, Oliver
Conventional Wisdom for Unconventional Times — Angus J. McLean, Oliver
DSPy: The End of Prompt Engineering - Kevin Madura, AlixPartners
DSPy is (really) All You Need - Kevin Madura, AlixPartners
The Complete DSPy Workshop - Kevin Madura, AlixPartners
Studio A/B test
Teaching AI to Find Real Vulnerabilities — Prof. David Brumley, Bugcrowd
Teaching AIs to Hack — Prof. David Brumley, Bugcrowd
"If an LLM has a Chrome zero-day, it would be able to hack nation-states." — Prof. David Brumley
Give Your Agent a Computer — Nico Albanese, Vercel
The Agent Gets a Computer: Building with AI SDK v6 — Nico Albanese, Vercel
Notion's Token Town — Sarah Sachs, Notion
Notion's Token Town — Sarah Sachs, Notion
We updated your video to use the winner
Full Workshop: Setting Yourself Up for Success —Jason Liu, OpenAI Codex
Full Workshop: Setting Yourself Up for Success —Jason Liu, OpenAI Codex
Mastering AI Pricing: Flexible & Agile Monetization — Mayank Pant, Stripe
How to Monetize your AI (without Killing Growth) — Mayank Pant, Stripe
Mastering AI Pricing — Mayank Pant, Stripe
Results with very similar performance
Perception Agents — Antje Barth, Amazon AGI Lab
Perception Agents — Antje Barth, Amazon AGI Lab
Every API Is a Tool for Agents - Matt Carey, Cloudflare
"The best MCP server is the one you didn't have to build" - Matt Carey, Cloudflare
MCP = Mega Context Problem - Matt Carey
Studio A/B test
Scaling the Next Paradigm of Heterogeneous Intelligence — Adrian Bertagnoli, Callosum
On Heterogeneous Intelligence — Adrian Bertagnoli
Mixture of Models, RLMs, & Hardware-Software Co-Evolution: The Case for Heterogeneous Intelligence
Studio A/B test
Simon Willison in conversation with Cat Wu & Thariq Shihipar, Anthropic
Claude Fable, Claude Tag, and Anthropic's Culture — Cat Wu & Thariq Shihipar ft Simon Willison
This Year in Claude: Fable, Tag and Anthropic's Culture — Cat Wu & Thariq Shihipar ft Simon Willison
Task Fidelity Scaling Laws — Kobie Crawdord, Snorkel
Same compute, 5x better results — Kobie Crawdord, Snorkel
Why your RL training data is producing noise — Kobie Crawdord, Snorkel
Studio A/B test
How Evals and Prompts Shape Agent Behavior — Preetika Bhateja & Daniel Bump, YouTube Ads
The YouTube Ads Model Whisperers on how Evals and Prompts Shape Agent Behavior
YouTube's Model Whisperers: How Evals & Prompts Shape Behavior — Preetika Bhateja & Daniel Bump
Results with very similar performance
Through the AI Fog: The Architectural Decision Agentic Security Depends On — Manoj Nair, Snyk
Through the AI Fog: The Architectural Decision Agentic Security Depends On — Manoj Nair, Snyk
Studio A/B test
Everything I Learned Training Frontier Small Models — Maxime Labonne, Liquid AI
Training Frontier Efficient Models — Maxime Labonne, Liquid AI
LFM 2.5: Architecture, Recipes, Post-Training — Maxime Labonne, Liquid AI
Studio A/B test
Everything You Need To Know About Agent Observability — Danny Gollapalli & Zubin Koticha, Raindrop
Humanity's Last Problem — Danny Gollapalli & Zubin Koticha, Raindrop
Full Workshop: AI Observability Beyond Evals — Danny Gollapalli & Zubin Koticha, Raindrop
We updated your video to use the winner
Memory Harnesses for Long-Running Research Agents — Stefania Druga, Sakana.ai
Memory Harnesses for Long-Running Research Agents — Stefania Druga, Sakana.ai
Results with very similar performance
How Forward Deployed Engineering is done at Decagon — Sunny Rekhi
How Forward Deployed Engineering is done at Decagon — Sunny Rekhi
Studio A/B test
Beyond Transcription: Building Voice AI That Understands Conversations — Hervé Bredin, pyannoteAI
Your benchmark uses headset audio. Your users don't. — Hervé Bredin, pyannoteAI
Speaker diarization: the part of your voice stack that isn't working — Hervé Bredin, pyannoteAI
Studio A/B test
Imagination Engineering — Eve Bouffard, Head of Design, Y Combinator
Imagination Engineering: "Live in the future and then build what's missing."
Y Combinator's Head of Design on Imagination Engineering
LLM codegen fails and how to stop 'em — Danilo Campos, PostHog
The PostHog Wizard: Lessons in AI onboarding — Danilo Campos
How to stop LLM Fails — Danilo Campos, PostHog
Recursive Model Improvement — Lee Robinson, Cursor, SpaceXAI
More Compute In ➤ Better Model Out — Lee Robinson, Cursor
"The model trains the next model" — Lee Robinson, Cursor, SpaceXAI
Studio A/B test
Spec-Driven Development: Sharpening your AI toolbox — Al Harris, Amazon Kiro
Spec-Driven Development: How to Control Your Agents — Al Harris, Amazon Kiro
Spec-Driven Development: Agentic Coding at FAANG Scale and Quality — Al Harris, Amazon Kiro
AgentCraft: Putting the Orc in Orchestration — Ido Salomon
Putting the Orc in Orchestration — Ido Salomon, AgentCraft
Studio A/B test
Build Agents That Run for Hours (Without Losing the Plot) — Ash Prabaker & Andrew Wilson, Anthropic
Anthropic Workshop: Build Agents That Run for Hours — Ash Prabaker & Andrew Wilson
Long-Running Agents: Evolution, Challenges, and Patterns — Ash Prabaker & Andrew Wilson, Anthropic
Studio A/B test
A Piece of Pi: Embedding The OpenClaw Coding Agent In Your Product — Matthias Luebken, Tavon
Embedding Agents In Your Product — Matthias Luebken, Tavon
Case Study: Embedding Pi In Your Product — Matthias Luebken, Tavon
Studio A/B test
Demand-Driven Context: A Methodology for Coherent Knowledge Bases Through Agent Failure
Let AI Agents Tell You What They Need — Raj Navakoti, IKEA Digital
Agents Learn Best From Failures — Raj Navakoti, IKEA Digital
Studio A/B test
20 days of compute vs 7 hours: rethinking what state-of-the-art means — Bertrand Charpentier, Pruna
Why leaderboard rankings disagree: and what to use instead — Bertrand Charpentier, Pruna AI
The top model loses 40% of its battles. Here's what that means. — Bertrand Charpentier, Pruna AI
You Might Not Need 50 Diffusion Steps — Ziv Ilan, Nvidia
Real Time Video Diffusion on a Single GPU — Ziv Ilan, Nvidia
From Fifty Steps to One: Distilling Diffusion Models — Ziv Ilan, Nvidia
Studio A/B test
Building Conversational Agents — Thor Schaeff and Philipp Schmid, Google DeepMind
[Full Workshop] Building Conversational Agents — Thor Schaeff and Philipp Schmid, Google DeepMind
Gemini Live: must-know APIs for Conversationsal Agents in 2026 — Thor Schaeff and Philipp Schmid
What we learned scaling MCPs to Enterprise — Karan Sampath, Anthropic
Gateways are All You Need — Karan Sampath, Anthropic
How Anthropic scales MCPs — Karan Sampath, Anthropic
Studio A/B test
Frontier AI and the Future of Intelligence — Raia Hadsell, VP of Research, Google DeepMind
How Google DeepMind is researching the next Frontier of AI for Gemini — Raia Hadsell, VP of Research
Frontier AI and the Future of Intelligence: Embeddings, GNNs, and World Models
Results with very similar performance
Bringing Continual Learning into Enterprises — Samuel Denton, Applied Compute
Bringing Continual Learning into Enterprises — Samuel Denton, Applied Compute
The Prime Intellect Stack — Will Brown, Prime Intellect
Modern Post-Training: A Deep Dive — Will Brown, Prime Intellect
Building Open Source Agentic Models — Will Brown, Prime Intellect
Studio A/B test
Engineering voice agents: Latency, quality, and scale — Rishabh Bhargava, Together AI
500ms is the cliff: engineering voice agents that don't lose callers — Rishabh Bhargava, Together AI
8B to 30B: the LLM sweet spot for production voice agents — Rishabh Bhargava, Together AI
Guide, Verify, Solve — Anirban Chatterjee, Sonar
Guide, Verify, Solve — Anirban Chatterjee, Sonar
Guide, Verify, Solve — Anirban Chatterjee, Sonar
Studio A/B test
2026 State of
"The biggest challenge in your stack? Evals, Evals, Evals" - 2026 State of
"Cost is now a first-class engineering constraint." — Barr Yaron, Amplify Partners
Evolution of agentic surfaces — Gagan Bhat & Isabella Kai He, Anthropic
Anthropic's Applied AI team on the Evolution of Agentic Surfaces
Claude Managed Agents and Evolution of Agentic Surfaces — Gagan Bhat & Isabella Kai He, Anthropic
Studio A/B test
OpenAI + @Temporalio : Building Durable, Production Ready Agents - Cornelia Davis, Temporal
Durable, Production Ready Agent Orchestration with OpenAI + Temporal - Cornelia Davis, @Temporalio
How OpenAI builds Production Ready Agents - Cornelia Davis, Temporal
How to Keep Shipping When You Walk Away from Your Desk — Zack Proser, WorkOS
Your Attention Is the Bottleneck, Not Your Agents — Zack Proser, WorkOS
The Hard Constraint: Building Around Human Attention in an Agent-First World — Zack Proser, WorkOS
Results with very similar performance
How Forward Deployed Engineering is done at Ramp — Leo Mehr
How Forward Deployed Engineering is done at Ramp — Leo Mehr
Results with very similar performance
The Messy Reality of Scale: Synthetic Data and Pre-Training — Marah Abdin & Robert McHardy, poolside
The Messy Reality of Scale: Synthetic Data and Pre-Training — Marah Abdin & Robert McHardy, poolside
We updated your video to use the winner
Don't Ship Skills Without Evals — Philipp Schmid, Google DeepMind
Don't Ship Skills Without Evals — Philipp Schmid, Google DeepMind
Building your own software factory — Eric Zakariasson, Cursor
How Cursor Codes — Eric Zakariasson, Cursor
How I Cursor — Eric Zakariasson, Cursor
Studio A/B test
Road to 5 Million Tokens: Breaking Barriers in Long Context Training — Max Ryabinin, Together AI
The Memory Stack Behind 5 Million Token Transformers — Max Ryabinin, Together AI
How Lovable self-improves every hour — Benjamin Verbeek, Lovable
The agent that files its own bug reports — Benjamin Verbeek, Lovable
How Lovable taught its agent to surface its own blockers — Benjamin Verbeek, Lovable
Studio A/B test
The Production AI Playbook: Deploying Agents at Enterprise Scale — Sandipan Bhaumik, Databricks
£85K Burned on a Failed PoC: What Actually Gets Agents to Production — Sandipan Bhaumik, Databricks
Studio A/B test
The Unreasonable Effectiveness of Separating the Task from the Model — Maxime Rivest & Isaac Miller
DSPy 4: The Unreasonable Effectiveness of Separating the Task from the Model
Identity for AI Agents - Patrick Riley & Carlos Galan, Auth0
Everything you need to know about AI Agent Security / Identity - Patrick Riley & Carlos Galan, Auth0
Authn for Agents - Patrick Riley & Carlos Galan, Auth0
Studio A/B test
From Writing Code to Designing Systems: How the Developer Role is Changing — Chris Noring, Microsoft
How the Developer Role is Changing — Chris Noring, Microsoft
So many AI Tools: When to use what? — Chris Noring, Microsoft
Studio A/B test
FLUX, Open Research, and the Future of Visual AI — Stephen Batifol, Black Forest Labs
Building Black Forest Labs: FLUX, Self-Flow, and World Models — Stephen Batifol
The Road to Visual Intelligence — Stephen Batifol
Studio A/B test
CI/CD Is Dead, Agents Need Continuous Compute and Computers — Hugo Santos and Madison Faulkner
CI/CD is dead — Hugo Santos, Namespace CEO and Madison Faulkner, NEA
Continuous Compute: Intents, Humans *Outside* the Loop and the Multiverse future of the SDLC
Studio A/B test
Special Topics in Kernels, RL, Reward Hacking in Agents — Daniel Han, Unsloth
Full Workshop: Reward Hacking in Agents — Daniel Han, Unsloth
Reward Hacking in Agents — Daniel Han, Unsloth
Studio A/B test
Collaborative
"What comes after the PR?" — Maggie Appleton, GitHub Next
Ace: a peek into the Collaborative
An AI Agent Became the #1 Contributor in OpenAI's Hiring Challenge — Zhengyao Jiang, Weco
Autoresearching to #1 in OpenAI Parameter Golf: 1 Agent vs 1000 Researchers — Zhengyao Jiang, Weco
How Autoresearch is changing ML research — Zhengyao Jiang, Weco
We updated your video to use the winner
Multiplayer agentic engineering — Arjun Singh, Superconductor
Multiplayer agentic engineering — Arjun Singh, Superconductor
Multiplayer agentic engineering — Arjun Singh, Superconductor
Cooking with Agents in VS Code — Liam Hampton, Microsoft
Local, Background, Cloud: 3 Agents in 1 IDE — Liam Hampton, Microsoft
Cooking with Agents — Liam Hampton, Microsoft
Studio A/B test
Two Roads to Durable Agents: Replay vs. Snapshot — Eric Allam, CEO, Trigger.dev
Durable Agents: Replay vs. Snapshot
Two Roads to Durable Agents — Eric Allam, CEO, Trigger.dev
Studio A/B test
The agent-ready web: Simplify user actions with WebMCP — Tara Agyemang, Google
The Complete Guide to WebMCP — Tara Agyemang, Google Chrome
WebMCP: How every website talks to agents — Tara Agyemang, Google Chrome
Studio A/B test
State of the Union: Why Local, Why Now — NVIDIA, Osmantic, Roboflow, EXO Labs, @matthew_berman
Local AI State of the Union: Why Local, Why Now — NVIDIA, Osmantic, Roboflow, EXO Labs, Matt Berman
Why Local, Why Now — NVIDIA, Osmantic, Roboflow, EXO Labs, @matthew_berman
Lessons from Scaling GitHub's Remote MCP Server — Sam Morrow, GitHub
GitHub's Remote MCP Server: Lessons from 1 Year in Production — Sam Morrow, GitHub
Scaling GitHub for your Agents — Sam Morrow, GitHub
Studio A/B test
Mergeable by default: Building the context engine to save time and tokens — Peter Werry, Unblocked
Mergeable by default — Peter Werry, Unblocked
Building the context engine to save time and tokens — Peter Werry, Unblocked
Studio A/B test
Prompt to Pipeline: Building with Google's Gen Media Stack — Paige & Guillaume, Google DeepMind
Gemini, Nano Banana, VO, & LIA: A Full Stack Gen Media Workshop — Paige & Guillaume, Google DeepMind
Build Real Apps with DeepMind's Model Suite — Paige Bailey & Guillaume Vernade, Google DeepMind
Studio A/B test
Lobster Trap: OpenClaw in Containers from Local to K8s and Back — Sally Ann O'Malley, Red Hat
Lobster Trap: OpenClaw in Containers — Sally Ann O'Malley, Red Hat
OpenClaw in Containers: The Lobster Trap — Sally Ann O'Malley, Red Hat
Studio A/B test
Your Agent Can Now Train Models — Merve Noyan, Hugging Face
Open Source Agents: Every must-know Hugging Face feature in 2026 — Merve Noyan
Self-Training Agents: Hermes Agent, HF Traces, Skills, MCP & Finetuning — Merve Noyan, Hugging Face
Why MLX — Prince Canuma, Neywa Labs
On-Device Intelligence with MLX — Prince Canuma, Neywa Labs
MLX: Enabling Generative Media on Apple Silicon — Prince Canuma, Neywa Labs
Fast Models Need Slow Developers — Sarah Chieng, Cerebras
Codex Spark: Fast Models Need Slow Developers — Sarah Chieng, Cerebras
1200 tok/s Fast Models Need Slow Developers — Sarah Chieng, Cerebras
Personalization in the Era of LLMs - Shivam Verma, Spotify
How Spotify models *you*: The Era of Generative Recommendations - Shivam Verma, Spotify
Spotify's LLM RecSys: 750M MAU x 100M tracks - Shivam Verma, Spotify
Studio A/B test
I Gave an AI Agent the Keys to My Life (Here's What Happened) — Radek Sienkiewicz (@velvetshark-com)
Making OpenClaw my Life Infrastructure — Radek Sienkiewicz (@velvetshark-com)
The 5 Jobs of my OpenClaw Life OS — Radek Sienkiewicz
The test completed with no winner
Claude for Long-Horizon Tasks — Lance Martin, Anthropic
Claude for Long-Horizon Tasks — Lance Martin, Anthropic
Claude for Long-Horizon Tasks — Lance Martin, Anthropic
Viktor: AI Coworker That Lives in Slack — Fryderyk Wiatrowski
Viktor: Your Last Hire — Fryderyk Wiatrowski
Building Viktor — Fryderyk Wiatrowski, CEO, Viktor
Results with very similar performance
"The engineer of the future is the person who is able to choose what is worth doing." — Addy Osmani
"The engineer of the future is the person who is able to choose what is worth doing." — Addy Osmani
"The engineer of the future is the person who is able to choose what is worth doing." — Addy Osmani
Results with very similar performance
How Web Data Infrastructure Powers the Next Generation of AI — Patricija Žemaitytė, Oxylabs
How Web Data Infrastructure Powers the Next Generation of AI — Patricija Žemaitytė, Oxylabs
6 Things to Know about AIE World's Fair 2026
Know before you Go: AIE World's Fair 2026
What's new at
Studio A/B test
On AI and Knowledge — Pablo Castro, Distinguished Engineer & CVP for AI Knowledge, Microsoft
Intrinsic + Extrinsic + Learned Knowledge — Pablo Castro, Distinguished Engineer & CVP, Microsoft
On AI and Knowledge
Voice AI: when is the "Her" moment? — Neil Zeghidour, CEO, Gradium AI
Speed, Cost, Full-Duplex, and Paralinguistic Understanding: What we need to build "Her"
Why Pipelines won't build Voice AGI — Neil Zeghidour, CEO, Gradium AI
The Missing Primitive for Agent Swarms — Lou Bichard, Ona
Coordination: The Missing Primitive for Agent Swarms — Lou Bichard, Ona
Ona: The Missing Primitive for Agent Swarms — Lou Bichard
Studio A/B test
The Friction is Your Judgment — Armin Ronacher & Cristina Poncela Cubeiro, Earendil
"The moments where you want to skip thinking are EXACTLY where thinking matters most" — Earendil
The Friction is Your Judgment — Armin Ronacher & Cristina Poncela Cubeiro, Earendil
Why TTS Models Now Look Like LLMs — Samuel Humeau, Mistral
How Modern Text-To-Speech Works — Samuel Humeau, Research Scientist, Mistral
How LLMs ate Speech — Samuel Humeau, Research Scientist, Mistral
Studio A/B test
How Google DeepMind Runs Agents at Scale — KP Sawhney & Ian Ballantyne, Google DeepMind
Inside Google's Agent Infrastructure: Quotas, Skills, and Deep Research — KP Sawhney, Ian Ballantyne
Building Agentic Systems Inside Google DeepMind — KP Sawhney & Ian Ballantyne, Google DeepMind
Results with very similar performance
Thinner Agents on a Smarter Substrate: The Ontology-based Semantic Layer — Emil Eifrem, Neo4j
Thinner Agents on a Smarter Substrate: The Ontology-based Semantic Layer — Emil Eifrem, Neo4j
Results with very similar performance
Improving Agents is a Data Mining Problem — Vivek Trivedy, LangChain
Improving Agents is a Data Mining Problem — Vivek Trivedy, LangChain
Results with very similar performance
Scaling up Continual Learning — Ronak Malde, Trajectory
Scaling up Continual Learning — Ronak Malde, Trajectory
Studio A/B test
Playground in Prod: Optimising Agents in Production Environments — Samuel Colvin, Pydantic
Agent Optimization with Pydantic AI: GEPA, Evals, Feedback Loops — Samuel Colvin, Pydantic
Full Workshop: Optimising Agents with GEPA, Logfire, Pydantic AI — Samuel Colvin, Pydantic
Studio A/B test
Give Your Chat Agent a Voice — Luke Harries, Head of Growth, ElevenLabs
ElevenLabs Voice Engine: Give Every Agent A Voice — Luke Harries
Voice is the Natural Medium — Luke Harries, ElevenLabs
The New Physics of Business — Garry Tan, Y Combinator
Every company should have a Brain — Garry Tan, Y Combinator
Never do one-off work — Garry Tan, Y Combinator
Studio A/B test
TLMs: Tiny LLMs and Agents on Edge Devices with LiteRT-LM — Cormac Brick, Google
Full Workshop: Using Tiny LLMs and Agents On Device — Cormac Brick, Google
[Workshop] When & How to use Tiny LLMs — Cormac Brick, Google
Studio A/B test
The Multi-Agent Architecture That Actually Ships — Luke Alvoeiro, Factory
Missions: The Multi-Agent Architecture That Actually Ships — Luke Alvoeiro, Factory
The Five types of Multi-Agent Architecture — Luke Alvoeiro, Factory
Open Models at Google DeepMind — Cassidy Hardin, Google DeepMind
Gemma 4 Deep Dive — Cassidy Hardin, Researcher, Google DeepMind
Gemma 4: Everything we did for Frontier On-Device Models — Cassidy Hardin, Google DeepMind
Studio A/B test
Most Enterprise Agentic Projects Are Doomed, Here's Why — Jess Grogan-Avignon & Jack Wang, Accenture
When MVPs take 2 weeks but Prod takes 12 months — Jess Grogan-Avignon & Jack Wang, Accenture
Accenture: Most Enterprise Agentic Projects Are Doomed — Jess Grogan-Avignon & Jack Wang,
Run Frontier AI at Home — Alex Cheema, EXO Labs
Full Workshop: Frontier AI at Home — Alex Cheema, EXO Labs
100x faster local inference for 1T param models across 4 Mac Studios — Alex Cheema, EXO Labs
Results with very similar performance
Computer-Use 2.0: Agents Just Got Multi-Cursor — Francesco Bonacci, Cua
Computer-Use 2.0: Agents Just Got Multi-Cursor — Francesco Bonacci, Cua
Computer-Use 2.0: Agents Just Got Multi-Cursor — Francesco Bonacci, Cua
We updated your video to use the winner
The Rise of CaaS: Context-as-a-Service for Agentic AI — Omer Primor, Bright Data
The Rise of CaaS: Context-as-a-Service for Agentic AI — Omer Primor, Bright Data
We updated your video to use the winner
Intelligence + Continual Learning = Expertise — Yu Su, NeoCognition
Intelligence + Continual Learning = Expertise — Yu Su, NeoCognition