Is a vector database enough for production AI agents? Reddit debates storage design
OkShirt9372 · reddit · 2026-09-28
A Reddit thread asks whether vector databases alone can support production AI agents. Agents must store conversation history, task state, tool results, and user preferences, with concurrent reads/writes across multiple agents and reliable task resumption — needs vector search doesn't cover, like exact lookups, metadata filters, transactional writes, and concurrent updates.
The poster lists open questions: isolated data copies for testing, combining vector + keyword + structured filters in one query, handling conflicting record updates, write-to-retrieval latency, and transactional vs. analytical workloads. Discussion covers whether to use one database for memory and operational state or split vectors, state, and transactional data across systems.
More from coding & agent
- OOD Labs launches Sentinel: agent-built real-time node graphs with live diffusion, $199.99 — PurzBeats · 2026-09-28
- mark_k: AI coding is approaching a step change — systems only AI can debug — mark_k · 2026-09-28
- Livestream: Sentinel, a Real-Time Node Graph Built Natively for Agents — PurzBeats · 2026-09-28
- Prasoon Pratham Flips on Vibe Coding: 'Coding With AI No Longer Sucks' — PrasoonPratham · 2026-09-28
- "MCP Was a Bad Idea": Devs Argue Smarter Models Made Most MCP Servers Obsolete — udmrzn · 2026-09-28
- Agent spends 510 hours decompiling Crimsonland, hits 96.4% match with 2 functions left — banteg · 2026-09-28