Skip to content

Search is only available in production builds. Try building and previewing the site to test it out locally.

Concepts

Postgram is a self-hosted private knowledge store for humans and agents. It keeps knowledge in typed entities and layers retrieval, graph context, and background processing on top of the same data model.

This page summarizes how Postgram is organized, what it stores, and what makes it useful for both humans and automation.

  1. Ingest content as typed entities (memory, person, project, task, interaction, document).
  2. Let Postgram enrich content in the background (chunking, embedding, optional extraction).
  3. Query with hybrid search to retrieve ranked results, including optional graph context.
  4. Use explicit or extracted relationships to navigate connected context.
  • Postgram is designed for private use by an individual or small team, not as a public multi-tenant SaaS platform.
  • Every public API is surfaced through REST, MCP, and CLI using the same core entity types.
  • For implementation details of each concept, see: Entity Model, Memory Roles, Hybrid Search, and Knowledge Graph.