# turbomem

> turbomem is an open-source (Apache-2.0) TypeScript library for embedded agent memory: LLM fact extraction, semantic vector search, tenant scoping, and pluggable storage backends.

turbomem runs inside your Node, Bun, browser, or edge process. No Python sidecar. No dedicated memory server. Install with `npm install turbomem`.

Storage adapters: PGlite + pgvector (default, local-first), sqlite-vec, Upstash Vector (edge/serverless), and Pinecone (managed cloud). Framework adapters include Vercel AI SDK and Mastra.

## Product

- [Home](https://turbomem.dev/): Overview, features, and how turbomem works
- [Templates](https://turbomem.dev/templates): Starter apps for building AI assistants with turbomem
- [npm package](https://www.npmjs.com/package/turbomem): Install and version info
- [GitHub](https://github.com/turbomem/turbomem): Source code, issues, and contributions
- [Demo](https://demo.turbomem.dev/): Interactive PGlite demo

## Documentation

- [Docs home](https://docs.turbomem.dev/): Full documentation site
- [Getting started](https://docs.turbomem.dev/guide/getting-started): Quickstart guide
- [CLI](https://docs.turbomem.dev/cli): Terminal commands for managing memories
- [Adapters](https://docs.turbomem.dev/adapters): Storage and framework adapter reference
- [API reference](https://docs.turbomem.dev/api): TypeScript API docs
- [Examples](https://docs.turbomem.dev/examples): Code examples

## Optional

- [Blog](https://blog.turbomem.dev/): Updates and articles
- [Contact](https://turbomem.dev/contact): Team contact and support
- [Privacy policy](https://turbomem.dev/privacy): Privacy policy
- [Terms of service](https://turbomem.dev/terms): Terms of service
- [Cloud waitlist](https://docs.google.com/forms/d/e/1FAIpQLSftLBnz2UK_4eVjc4uf_Leg4HyqtYN2m6ppE_rpU5R-rT7Lig/viewform): Sign up for managed turbomem Cloud (coming soon)