Hi HN — I’m one of the builders behind SochDB.

SochDB is a high-performance embedded, ACID-compliant vector database purpose-built for AI agents and memory

We built it after repeatedly running into complexity around SQL + vector DB + context/memory when building LLM and agent apps.

SochDB is embedded and local-first by design. The goal is to reduce glue code and make reasoning about state simpler.

Happy to answer questions about: - embedded vs cloud trade-offs - SQL + vector data modeling - where this fits (and doesn’t)

Open to all feedback, including critical takes.