LettaBot

LettaBot is a stateful agent framework that uses tiered memory management to give LLMs long-term context and autonomous tool-use capabilities.

LettaBot (built on the Letta framework) solves the statelessness of standard LLMs by implementing a MemGPT-inspired architecture. It manages information across three layers: a fixed context window, a searchable archival database, and a persistent recall storage. This setup allows the bot to retain user preferences across sessions, execute multi-step workflows without human intervention, and self-edit its own memory. Developers deploy it via the Letta API to build agents that actually remember past interactions and operate reliably in production environments.

https://letta.com