Light
Dark

The human brain is a sponge. Today’s AI brains are brittle and rigid. At Letta, we’re building self-improving artificial intelligence: creating agents that continually learn from experience and adapt over time.

We’re assembling a world-class team of researchers and engineers to solve AI’s hardest problem: making machines that can reason, remember, and learn the way humans do.

Solving Self-Improving Superintelligence

Our team is focused on understanding the fundamental limitations of LLM-driven intelligence, and overcoming these limitations through foundational research in areas such as memory, continual learning, and scaling test-time compute.

Featured papers

Sleep-time Compute: Beyond Inference Scaling at Test-time

We introduce sleep-time compute, which allows models to think offline while they "sleep" by continually processing their raw context.

Read the paper

MemGPT: Towards LLMs as Operating Systems

Read the paper

Lost in the Middle: How Language Models use Long Context

Read the paper

Few-Shot Adaptation for Parsing Contextual Utterances with LLMs

Read the paper

RALF: Accuracy-Aware Scheduling for Feature Store Maintenance

Read the paper

Join our team

Founded by the creators of MemGPT from UC Berkeley’s Sky Computing Lab (the birthplace of Spark and Ray). Backed by Jeff Dean, Clem Delangue, and pioneers across AI infrastructure.

Our agents already power production systems at companies like 11x and Bilt Rewards, learning and improving every day.
About usSee open positions

Foundational Research

Our team is pursuing foundational research in AI memory (parametric and non-parametric representations), learning, and self-improvement.

Open AI

Letta is an open AI company. We publish and share our research, and contribute our advances directly into our open source software.

Extreme Impact

In our small, high-agency team, every researcher drives major advances that immediately improve the capabilities of AI agents today.

AI Developers Love

We obsess over making AI that is delightful to work with, while pushing the boundaries of what's possible with LLMs.