Turn a goal into a workflow that runs on your schedule. Subagents work in parallel, actions you choose wait for approval, and every run shows its cost and stays inside the limits you set.
Learn more
See what each run costs, split into model, subagent, tool, and system spend. Set spend, time, and token limits per run before it starts, and get a warning before a run reaches them.
Learn more
Auto sends each request to the model tier it needs: efficient for quick asks, fast for everyday work, and frontier for hard, multi-step problems. Pick a model yourself whenever you want.
Learn more
Connect your own OpenAI, xAI, or OpenRouter key, or your ChatGPT or X subscription, and matching model calls run on your account first. Run turns on the built-in runtime, Claude Code, Codex, or Pi.
Learn more