Neon adds hybrid vector and full-text search for agent workflows
matei_zaharia · x · 2026-07-21
Neon says it has made Lakebase Search available, adding hybrid vector and full-text search for agent workflows.
The post explains why the team built lakebasetext and argues that Postgres' default BM25 approach is not enough for this use case. The accompanying diagram contrasts tsrank in GIN with Lakebase BM25, highlighting three differences:
- BM25 uses corpus-wide IDF, so rare terms score higher.
- It adds length normalization so long documents are not unfairly boosted.
- Scores remain more stable as the corpus grows.
The key point is that retrieval for agents needs more than simple term frequency ranking; it needs a search layer designed for scale and relevance.
Related event: Neon Launches Lakebase Search for AI Agent Workflows(2 posts)→
More from coding & agent
- Tenable and AWS launch a Black Hat build event for open-source security agents and MCP servers — Dave_Maynor · 2026-07-22
- Codex helps build Valdiluce, an open-world game with climbing, gliding and gondolas — Dimillian · 2026-07-22
- HeyGen adds a media-sourcing skill for coding agents with 75k images and 10k tracks — HeyGen · 2026-07-22
- Agent search bottlenecks are now about variance, not raw latency — rohanpaul_ai · 2026-07-22
- LangSmith adds tracing for Pipecat, LiveKit, OpenAI Realtime, and Gemini Live — LangChain · 2026-07-22
- An MCP server signs every AI agent tool call into a verifiable Merkle chain — Funky_Chicken_22 · 2026-07-22