Skip to content

Bob van Luijt

Weaviate

  • 2 episodes on Chain of Thought
  • 5 topics covered
  • 2 full transcripts
  • Nov 2024 first on the show
Bob van Luijt

Bob van Luijt is the co-founder and CEO of Weaviate, the open-source vector database. Across two Chain of Thought appearances he helped define agentic architecture and then returned for the Season 2 premiere to dig into how RAG and AI infrastructure are evolving, arguing that this is the time to start building.

Cover art for AI Infrastructure & the Evolution of RAG | Weaviate's Bob van Luijt

Latest episode with Bob van Luijt · EP 8

AI Infrastructure & the Evolution of RAG | Weaviate's Bob van Luijt

Bob van Luijt, Weaviate ·

Show notes & transcript → All listening options →

Episodes

What the conversations cover

These notes summarize the episodes featuring Bob van Luijt, including other speakers in panel conversations. Follow a source link to read the discussion in context.

EP 8 AI Infrastructure & the Evolution of RAG | Weaviate's Bob van Luijt

  • Weaviate already had what we today call RAG — they called it “generative search.” When the research-paper term caught on, they simply adopted the language: “in the product itself, nothing changed. It was just how we talked about it.” Episode takeaways
  • Bob frames the arc in three steps: RAG is “one directional” (query → retrieve → generate), agents add a feedback loop, and generative feedback loops — his favorite — let the agent “put stuff back inside the database.” Episode takeaways
  • The killer use case is master data management. Pre-Weaviate, consulting for a global candy manufacturer, Bob watched humans hand-reconcile messy data (Fahrenheit vs. Celsius, British vs. American spelling) — losing, because “data is coming in faster than these people could handle it.” Generative feedback loops let a model do that reconciliation instead. Episode takeaways

EP 2 Got Agents? Agentic Workflows & Architecture | Weaviate, Unstructured & CrewAI

  • João Moura (CrewAI) draws the line at agency: a fixed if-this-then-that flow isn’t an agent. An agent “controls the flow of the application,” using an LLM “as a brain” to reason about what tools to use, when to use them, and how to format the answer — “something that has agency.” Episode takeaways
  • Bob van Luijt (Weaviate) argues data is still the bottleneck, not models. A CIO told him: “it’s all great what you guys are building over there… but I still have like 15 ERP systems and I can’t, you know, I can’t make heads or tails of it.” Low data quality and not getting to insights remains “the number one problem.” Episode takeaways
  • João Moura (CrewAI) warns agent architectures start simple but explode in production: once inside an enterprise you bolt on a caching layer, a memory layer (long-term memory of what the agent has done and learned), guardrails, validation, security, and distribution — and the complexity “can get…to skyrocket a little bit.” Episode takeaways

In their words

It's brought us very far, but now we need to start to bring in loops, right? So what if you have more questions? What if we need to query multiple times? And that is something we at We've Hit, we put a lot of work in that. So we'll probably get to that later, but the feedback loops to get might be multiple sources of information, do multiple queries from the database.

Bob van Luijt Got Agents? Agentic Workflows & Architecture | Weaviate, Unstructured & CrewAI · 16:49

Transcript excerpts; ellipses mark omitted words. Some automatically selected quotes have filler words removed.