He’s Building the Document Layer for AI Agents | Jerry Liu, LlamaIndex
🔗 Link do vídeo: https://www.youtube.com/watch?v=QR_yEI8mNGo
🆔 ID do vídeo: QR_yEI8mNGo
📅 Publicado em: 2026-08-31T16:00:36Z
📺 Canal: Composio
⏱️ Duração (ISO): PT43M25S
⏱️ Duração formatada: 00:43:25
📊 Estatísticas:
– Views: 155
– Likes: 2
– Comentários: 0
🏷️ Tags:
Jerry Liu built LlamaIndex — the tool half the industry used for RAG. Then he called RAG a hack, about two years before the field agreed. In 2026, naive RAG is widely considered a liability, and the field has landed on exactly the problem he pointed at.
Here's that problem: 90% of the world's information is locked inside PDFs, PowerPoints, and Word documents. The models are already smarter than humans. But until an agent can actually read and reason over all that, it just sits still. RAG was the hack everyone reached for to bridge the gap — chunk, embed, retrieve, done — and it falls apart on anything complex.
What he's building instead is the layer underneath: document infrastructure for AI agents.
LlamaIndex started with one commit and one tweet in November 2022 — before ChatGPT, before RAG was even a term. It's now at 25M+ monthly downloads with Fortune 50 customers.
In this episode:
→ Why the person most associated with RAG called it a hack — two years before the field agreed
→ Why the bottleneck was never model intelligence — it's context
→ Why 90% of enterprise knowledge is trapped in documents, and what that means for agents
→ Why even MCP doesn't fully solve context — and what's still unsolved
→ How you get agents to follow a company's culture, not just its instructions
→ The origin: feeding GPT-3 private data inside a 4,000-token window, before RAG existed
→ What he'd build if he started LlamaIndex today
Follow Jerry: @jerryjliu0
⏱️ TIMESTAMPS
00:00 The Context Bottleneck
02:00 One Commit, One Tweet: The Origin
05:00 Feeding GPT-3 Your Private Data, Pre-RAG
08:00 From Quora to Uber's Self-Driving Lab
11:00 Meeting His Co-Founder at Uber ATG
14:00 Document Infrastructure for AI Agents
18:00 "RAG Is a Hack" — What He Meant
22:00 Why Production Search Is Still Unsolved
24:00 How a Research Background Changes How You Build
33:00 The Open Problem: Context Across Sources
38:00 Specs, Cultural Context, and Underspecified Tasks
40:00 What He'd Build Differently Today
This is a Composio "Agents at Work" podcast, where I sit down with founders building the next leap of AI.
🎧 Subscribe for new episodes
📩 Listen on Spotify: https://open.spotify.com/show/4fxtM0FC4FxTR4l9NRgWsj?si=d05940c466154f1e
🐦 Follow Composio: @composio
#AI #Agents #MachineLearning #Startups #RAG