AutoIndex treats RAG indexing as code optimization — LLM agents write your chunking programs
CShorten30 · x · 2026-09-07
The Weaviate Podcast #143 features Sam O'Nuallain (UMass Amherst) on AutoIndex, which reframes corpus indexing as a code optimization problem.
- Core idea: chunking isn't a config you tune — an analysis agent and a code agent loop together to write Python "representation programs" that chunk, enrich, and reorganize your corpus.
- Every hypothesis must prove validation lift before it survives.
- Biggest lesson: "did recall go up?" is useless feedback. Giving the analysis agent tools to investigate why a gold document ranked low is what made the system work.
Related event: UMass AutoIndex Lets LLMs Write Code to Optimize Indexing(3 posts)→
More from coding & agent
- Your Local AI Agent Harness Can Still Be a Landlord: Self-Hosted Doesn't Mean Safe — alex_verem · 2026-09-08
- Astra plans, Sol organizes, Luna executes: a solid three-agent workflow — tobowers · 2026-09-08
- One phone prompt, fully automated: agent finds video in email, cuts TikTok-ready reels, uploads to YouTube — brandon_galang · 2026-09-08
- Dev shares how to drive Blender with Codex locally for AI 3D modeling — bennash · 2026-09-08
- astra + Blender MCP near one-shots a 3D house model from four photos — sidahuj · 2026-09-08
- Under-2-second sandbox boot makes latency a non-issue; VM reliability and snapshots matter more — sh_reya · 2026-09-08