← Directory
Agent frameworks
Lettuce (Letta)
Framework for building long-context LLM agents with persistent memory and state.
Overview
Letta specializes in agents that remember context over long periods. It implements a memory architecture that mimics human recall, allowing agents to maintain personality and history across sessions without relying solely on context windows.
Best for
- Long-running agents
- Character AI
Trade-offs
vs LangChain Memory
Letta is purpose-built for deep memory; LangChain memory is generic and often shallow.
What XeroHack pre-wires
When you pick Lettuce (Letta) in the interview, the engine emits these files into your scaffold:
- Agent config
- Memory store
Coupon
No active coupon for this tool right now. Sign in to be notified when one is available.
Sign in