# Letta

> Letta is a San Francisco AI research lab and platform for building stateful agents - AI that remembers, learns, and improves itself over time. Spun out of UC Berkeley's Sky Computing Lab from the viral MemGPT project, Letta gives developers an open framework, API, and cloud service where agent state (memories, messages, reasoning, tool calls) is persisted in a database rather than lost when it falls out of the context window. The company positions itself as the open alternative for developers who want to build agents that behave less like stateless chatbots and more like continuously learning digital collaborators.

- **Founded:** 2023
- **Headquarters:** San Francisco, California, United States
- **Founders:** Charles Packer (Co-founder & CEO), Sarah Wooders (Co-founder & CTO)
- **Team size:** ~16 employees
- **Products:** Letta (open-source framework), Letta Cloud, Letta API & SDKs, Letta Code, MemGPT (research foundation)
- **Notable:** MemGPT paper went viral on Hacker News in October 2023 before its official release, Open-source repo grew to 23,000+ GitHub stars and 2,500+ forks, Raised $10M seed at a $70M valuation led by Felicis with angels including Jeff Dean and Clem Delangue

## Products & services

- **Letta (open-source framework)** — Apache-2.0 platform for building stateful agents whose memory, messages, reasoning, and tool calls are persisted in a database so state is never lost even when evicted from the context window.
- **Letta Cloud** — Hosted service for developers to build, deploy, and scale stateful agents with advanced memory systems without managing infrastructure.
- **Letta API & SDKs** — A full-featured agents REST API with Python and TypeScript SDKs for integrating stateful, memory-enabled agents into applications.
- **Letta Code** — An open-source, deeply personalized coding agent harness that learns from experience and improves with use, built on Letta's research in AI memory and continual learning.
- **MemGPT (research foundation)** — The original self-editing memory architecture for LLMs - an 'operating system for LLMs' that pages memory in and out of the context window - that Letta commercializes and extends.

## Achievements

- MemGPT paper went viral on Hacker News in October 2023 before its official release
- Open-source repo grew to 23,000+ GitHub stars and 2,500+ forks
- Raised $10M seed at a $70M valuation led by Felicis with angels including Jeff Dean and Clem Delangue
- Introduced 'sleep-time compute,' shifting reasoning to idle time and cutting test-time workload up to 5x on benchmarks like GSM-Symbolic and AIME without hurting accuracy
- Introduced git-based context repositories (2026), bringing versioning to agent memory

## Latest updates

- **2024-09** — Emerged from stealth with a $10M seed round led by Felicis at a $70M post-money valuation.
- **2025-04** — Published 'Sleep-time Compute,' a new scaling direction where agents reason during idle time; shown to cut test-time compute up to 5x on math benchmarks.
- **2025-12** — Released research on 'Continual Learning in Token Space' - learning by updating context rather than model weights.
- **2026-02** — Introduced Context Repositories: git-based memory, rebuilding agent memory around programmatic context management and versioning.

## Links

- Website: https://www.letta.com
- LinkedIn: https://www.linkedin.com/company/letta-ai
- Twitter/X: https://twitter.com/letta_ai
- GitHub: https://github.com/letta-ai

---

Profile page: https://yespress.io/letta
Published by YesPress — https://yespress.io
Last updated: 2026-08-04