Introducing tip-of-the-tongue search for agent conversations Find and resume past sessions, even when all you have is a vague memory Try with `/search`
Letta
Research Services
San Francisco, CA 5,262 followers
The platform for building stateful agents
About us
Today's AI agents struggle with fragmented context and limited memory. Just as computers needed operating systems to manage resources, your agents need intelligent context management to unlock their full potential. Letta is a complete operating system for building stateful agents AI agents. Letta handles state persistence, context compilation, and intelligent resource allocation, so you can focus on building applications powered by agents that actually learn. Letta was founded by the creators of MemGPT, which introduced memory (or context) management for AI agents. The company was spun out of the UC Berkeley Sky Computing research lab, and is advised by Ion Stoica (co-founder of Databricks, Anyscale) and funded by top-tier investors.
- Website
-
https://letta.com
External link for Letta
- Industry
- Research Services
- Company size
- 11-50 employees
- Headquarters
- San Francisco, CA
- Type
- Privately Held
- Founded
- 2024
- Specialties
- artificial intelligence, ai agents, and machine learning
Locations
-
Primary
Get directions
San Francisco, CA, US
Employees at Letta
Updates
-
Introducing Context Repositories: git-tracked files for storing agent context. Agents can now write scripts and spawn memory subagents to programmatically restructure prior context and learn in token-space. We've also added built-in skills & subagents leveraging Context Repositories for Letta Code's memory: - `/init` (learn from old CC/Codex sessions) - reflection (learn from recent history) - defragmentation (reorganize and restructure memory over time) To enable Context Repositories, you can run `/memfs enable` and use `/memory` to view your agent's memory. Full blogpost linked in comments!
-
Letta Code supports background subagents. Features: - Parallel in one tool call - Non-blocking, primary agent continues - Completion notifications + status polling - The primary can ask subagents questions after task completion Just ask your agent to use background subagents. Learn more about background subagents here: https://lnkd.in/gwGNnhu8
-
Today's office hour recording is available. - Opus 4.6 benchmarks - Agents forking themselves - Live agent-edited Obsidian - Letta Cowork - Lettabot skills
Letta Office Hours: Opus 4.6, Lettabot Updates, Agent File Directory, and More
https://www.youtube.com/
-
LettaBot is your personal AI assistant that remembers everything across Telegram, Slack, WhatsApp, Discord, and Signal. LettaBot allows your agents to write code, use skills, and work autonomously. Learn more in our YouTube video:
LettaBot: The AI assistant that remembers everything
https://www.youtube.com/
-
Not a fan of terminals? You can now use Letta Code through a UI, with our fork of Agent Cowork (an OSS alternative to Anthropic Cowork). Co-work locally alongside agents with: - persistent memory - long-term learning - portability across model providers See the OSS repo and install at: https://lnkd.in/gunBAMtZ
-
The Letta Office Hour recording is available. This week we dive into LettaBot. LettaBot deploys your Letta agent locally and lets you communicate with it from anywhere via Signal, Telegram, WhatsApp, or Slack. Also -- a quick Claude Subconscious demo. Link 👇️ Watch this:
Letta Office Hours: Introducing LettaBot +Claude Subconscious Demo
https://www.youtube.com/
-
Introducing LettaBot 👾: A proactive personal assistant with perpetual memory, living on your computer. - Import skills from MoltHub (previously ClawdHub) & Vercel's skills.sh - Message from Telegram Messenger, WhatsApp, Slack, Signal - Built on the Letta Code SDK https://lnkd.in/gHN_zS9t
-
The recording for the Letta Office Hour is now available:
Letta Office Hours: Letta Chat, GitHub Action, Letta Code, and more!
https://www.youtube.com/