How it works

One big task in. The lead agent splits it up, then focuses.

A request that would bury a single agent gets broken apart, so no one agent has to hold all of it at once. The product comparison you can run in the demo is the whole idea.

The delegation story, in three steps.

Ask the team to compare two products. Watch the lead agent hand each side to its own specialist, in parallel, then come back with a clear verdict.

You ask for something big

A request that has a lot going on underneath it.

"Compare these two products."

Specialists take the hard parts

The lead agent hands each piece to a specialist that does only that, in parallel.

Each product's research (find reliable sources, pull the facts) goes to its own specialist.

The lead agent focuses on the answer

Each specialist reports back with a clear, finished answer, not raw work to sift through. The lead agent only has to weigh those answers, which keeps it free to focus on its own task: the actual comparison.

A clear verdict, because no single agent had to hold all of it at once.

Composition

Any agent can be another agent's tool.

Other AI agent frameworks treat subagents as ephemeral background workers with hard-coded depth limits and reduced tool access. We treat every agent as a first-class composable tool, with the same identity and the same tools, anywhere in a tree.

Once you configure an agent, it shows up to its parent as just another tool to call. No depth limit. No circular references. Any agent you build becomes reusable across every workflow that needs it.

No depth limit

Nest trees as deep as the task demands. Manager, research, browser, parser, tool is a real shape.

Reusable

Write a browsing specialist once, plug it into any other agent, at any depth.

Composable

Create any hierarchy tree to suit your workflow. As many hierarchies as you need.

Runtime

Every workflow runs in its own container.

Each top-level agent gets a container with exactly the dependencies it and its subagent tree needs, and nothing more. Workflows stay reproducible and easy to reset.

No more "my agent's environment is broken" headaches. You control the environment and can recreate it from scratch, based on the configuration. A browsing agent's Playwright install never leaks into your email agent's environment, each agent can only touch what its container's mounts grant it, and deleting an agent takes its container and files with it.

Hosted or self-host