Letta
Stateful agent platform with persistent long-term memory across sessions
View on GitHubWhat it solves
The memory window problem: agents that remember what users told them weeks ago without the developer having to build a retrieval pipeline or manage context stuffing by hand.
Overview
Formerly MemGPT, Letta is a UC Berkeley research project turned production platform for agents that remember things across conversations. The core idea: agents manage their own memory context window, deciding what to move into long-term storage and when to retrieve it, rather than the developer managing chunks manually. Ships a REST API, Python and TypeScript SDKs, and a hosted cloud tier.
Key facts
- Language
- Python
- License
- Apache 2.0
- Maturity
- Stable
- Maintainer
- letta-ai
- Reviewed
- 2026-06-11
Where to start
The examples directory shows the memory self-editing loop in a minimal chat scenario, watch the agent compact and recall its own context.
examples/Technologies
More in this domain
LangGraph
Stateful multi-agent orchestration for LLM applications
The state management problem in multi-agent systems: how you build agents that pass context across turns without losing the execution thread.
AutoGen
Multi-agent conversation framework from Microsoft Research
How to coordinate multiple specialized agents toward a shared goal without writing low-level orchestration logic by hand.
CrewAI
Role-based multi-agent orchestration for collaborative task execution
The boilerplate problem in multi-agent setup: defining roles, delegation rules, and task routing without writing a custom orchestration graph.