Light
Dark
LLM Agents
LLM Agents
LLM Agents
LLM Agents
LLM Agents

Building the Next Generation of Agentic Systems

Everyone's focused on making better processors, while we're building the operating system of the future. LLMs have unlocked a new generation of intelligent machines, but the model itself is just one piece of a complete agentic system.

At Letta, we're building agents that can truly reason, plan, learn, and remember. We see a future where AI can think more like we do, and we're creating the architecture to make that possible.

Research at Letta

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. Our mission is to build agentic systems that can reason, plan, and remember in ways current models cannot.

Featured papers

MemGPT: Towards LLMs as Operating Systems

Teaching language models to manage their own memory for infinite context.

Read the paper

Cloudcast: High-Throughput, Cost-Aware Overlay Multicast in the Cloud

Read the paper

Hindsight Task Relabelling: Experience Replay for Sparse Reward Meta-RL

Read the paper

Differentiable Neural Network Architecture Search

Read the paper

RALF: Accuracy-Aware Scheduling for Feature Store Maintenance

Read the paper

Skyplane: Optimizing Transfer Cost and Throughput using Cloud-Aware Overlays

Read the paper

Assessing Generalization in Deep Reinforcement Learning

Read the paper

Few-Shot Adaptation for Parsing Contextual Utterances with LLMs

Read the paper

Lost in the Middle: How Language Models use Long Context

Read the paper

Join Our Team

Work with a tight-knit, world-class research team led by PhDs in AI and systems from UC Berkeley’s BAIR and Sky research labs. Letta offers what larger companies cannot: the freedom to publish your work openly, the opportunity to design the architecture of the world’s most advanced AI systems, and the ability to drive major decisions rather than incremental improvements.
View teamSee open positions

Foundational Research

We pursue breakthroughs in agent architecture, memory systems, and reasoning that tackle the core limitations of today’s models to enable truly intelligent systems.

Open AI

At Letta we believe in open AI. We publish our research openly 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

Our research directly translates into advancing agent capabilities for developers building AI applications. We obsess over making our systems delightful to work with while pushing the boundaries of what's possible with LLMs.