LocalRecall
A local, no-GPU RESTful memory and knowledge-base layer for AI agents and chatbots, storing files and text in a vector database behind a web UI.
LocalRecall is a lightweight RESTful API for managing knowledge bases and files stored in vector databases, requiring no GPU, internet connection or cloud service. It gives agents and chatbots a simple, generic abstraction for retrieval-augmented generation and for long-term and short-term memory.
It ships with two vector database engines: Chromem, a local file-based store used by default, and PostgreSQL with TimescaleDB, pgvector and pgvectorscale for hybrid BM25-plus-vector search in production. A web UI supports uploading and managing files, including raw text, in Markdown, plain text and PDF formats, and it integrates with LocalAI, LocalAGI and other agent frameworks.
It is written in Go, requires Go 1.16 or higher to build, and can run standalone or in Docker, serving its web UI on port 8080 by default.