← back
Effective agent design patterns in production — Laurie Voss, LlamaIndex
Takeaway
Build agents around LLMs' real strength — compressing messy text into structure — and lean on frameworks like LlamaIndex to skip boilerplate.
Summary
- Laurie Voss (LlamaIndex DevRel) defines an agent as semi-autonomous software that uses tools to hit goals without explicit step-by-step direction.
- Key principle: LLMs excel at 'turn a large body of text into a smaller one' (summarization, contract interpretation, invoice processing) — avoid asking them to expand small prompts into large outputs.
- LlamaIndex stack: framework (Python/TS), LlamaParse for PDF/Word/PPT, LlamaCloud (paid hosted retrieval), LlamaHub for 400+ model and DB integrations.
- Encourages going beyond chatbots — embed LLMs in existing software to convert messy unstructured input into structured data for traditional pipelines.
- Discusses design patterns for production agents in his remaining time — including RAG as foundation and patterns to improve agent reliability.
agentsllamaindexrag
Original description
At LlamaIndex we see a lot of agents built every day, and we've got a sense of what works and what doesn't. We've distilled those learnings down into a series of patterns and best practices for building real-world, production agents, and we're here to share them. You'll learn patterns for applying structure and guidance to famously nondeterministic LLMs and get concrete instruction on how to implement them. About Laurie Voss Laurie has been a web developer for 27 years, and along the way he co-founded npm, Inc.. He cares passionately about making technology accessible to everyone by demystifying complex technology topics. Recorded at the AI Engineer World's Fair in San Francisco. Stay up to date on our upcoming events and content by joining our newsletter here: https://www.ai.engineer/newsletter