Guide & Resources for Hindsight

Hindsight gives AI agents persistent memory, store facts through an API or MCP server, and recall them later with semantic search powered by embeddings. Everything needed to extract and store memories runs locally on your own server: the embedding model, the reranker, and the Postgres + pgvector database that holds memory banks and their vector embeddings. The only external dependency is the LLM provider you configure, used to extract and reason over what gets remembered. Access is controlled by a single server-side access key rather than user accounts, the same key protects the web console, the REST API and the built-in MCP server, so any MCP-compatible agent client can connect directly. Python and TypeScript client libraries are available for integrating memory storage and recall into your own applications.

Start using Hindsight with Elestio. Here are some guides and resources to help you get started.