letta-ai/letta

Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.

Letta – Stateful AI Agents with Memory

What it is

  • Letta (formerly MemGPT) is an open‑source framework for building AI agents that retain state and memory across interactions. It lets developers create agents that can learn, improve, and keep a persistent identity.

Core components

  • Agent harness & runtime – the core engine that manages memory, reasoning, and tool usage.
  • Interactive terminal UI – a command‑line interface (letta) for chatting with an agent locally.
  • App Server – a self‑hosted HTTP server (letta server) that exposes agents via REST/WebSocket, enabling integration with other services.
  • Channels – built‑in connectors for Slack, Telegram, Discord, and custom channels, so agents can be reached from popular messaging platforms.
  • Letta Agent SDK – a TypeScript library that lets developers embed agents directly into their own applications.
  • Letta Cloud – optional hosted service that stores an agent’s memory, identity, and conversation history, making the same agent available across devices.

Getting started

# Install the CLI globally via npm
npm install -g @letta-ai/letta-code

# Open a local terminal chat UI
letta

# Or run the server for API access
letta server

How you can use it

  • Run a personal chatbot in your terminal.
  • Deploy a self‑hosted agent that your team can talk to via Slack/Discord.
  • Embed a memory‑backed assistant into a TypeScript web or desktop app using the SDK.
  • Store and sync an agent’s long‑term memory with Letta Cloud for continuity across sessions.

Documentation & resources

Project status

  • Actively developed under the letta-ai/letta-code repository.
  • The original V1 code lives in the archive branch for reproducibility, but new features and bug fixes are made in the main branch.

Who it’s for

  • Developers who want to prototype or ship AI agents that need persistent context.
  • Teams looking to integrate conversational AI into existing workflows (e.g., Slack bots) without building memory management from scratch.

All information above is taken directly from the repository’s README.

Related

  • Project
    letta-ai/letta-code

    Letta Code is a stateful agent framework that enables AI agents to maintain long-term memory, learn new skills, and evolve their own behavior over time through self-reflection and persistent state management.

  • Dispatch
    Exploring Memory Systems for AI Agents

    An exploration of the current landscape of memory systems for AI agents, focusing on thetextual analysis of how developers are approaching thetextual analysis of state management and context window managements

  • Project
    sopaco/cortex-mem

    A Rust-based long-term memory framework for AI agents that uses a three-tier hierarchical architecture and a virtual filesystem to provide persistent, context-aware interactions.

  • Project
    vercel-labs/personal-agent-template

    An open-source template for deploying a personal AI agent with persistent memory and integrations for Slack, GitHub, and Linear.

  • Project
    HKUDS/nanobot

    An ultra-lightweight, self-hosted personal AI agent framework that integrates multiple LLM providers, long-term memory, and various chat apps into a single runtime.