Skip to main content
Autonomy is a foundation to build and scale products that use teams of deep work agents to autonomously perform long and complex tasks. Each agent can plan, take notes on a filesystem, curate context on every turn, and delegate work to parallel sub-agents.

Autonomy Framework

The Autonomy Framework is an open-source python framework to develop apps that use thousands of collaborating autonomous agents to quickly and reliably get things done. Your apps can easily create deep work agents that excel at long and complex tasks. Each agent can access a dedicated filesystem to save plans, take notes, and record memories. Agents can create many sub-agents that concentrate on sub-tasks. Parallel agents, each focused on a small part, achieve the overall goal faster. They are also more successful because the context that they provide to their language model contains much less noise.

Autonomy Computer

The Autonomy Computer is a cloud-based runtime to deploy, connect, and scale apps built using the Autonomy Framework. It provides infrastructure to run highly scalable and secure autonomous AI products. Using a simple YAML config, you can spin up a dedicated cluster of nodes in seconds. These nodes are automatically connected together using high-throughput, encrypted messaging channels. Agents invoke tools to take actions and gather data. The Autonomy Computer can provision tools in containers right alongside the nodes that are running your agents. These tools can be Python functions, simple binaries, or MCP servers. You can also create private links to enterprise data. The runtime is designed for multi-tenant, streaming workloads. Agents run as stateful concurrent actors that asynchronously send and receive messages. This enables effortless horizontal scaling and fan-out through powerful patterns like distributed agentic map-reduce.

Autonomy Command

The Autonomy Command is a command-line tool to build apps using the Autonomy Framework and deploy them to the Autonomy Computer. It packages source code into container images and runs them in your cluster.