Mercury

Mercury is a canvas for running multiple AI agents alongside a human team. You spin up coordinated agent teams, mix agents from Claude, OpenAI, and other frameworks, and set policies and approvals so work stays under human control. Agents connect to more than 800 tools through Composio, and you interact with them through a web UI, Slack, and iMessage rather than a separate dashboard. The result is task delegation that persists across long, multi-step workflows: agents pick up jobs, hand off to each other, and route decisions back to people when a policy requires sign-off. Mercury is no-code and real-time, aimed at teams that want to put agents to work on real projects without building orchestration plumbing themselves. It suits product and operations groups experimenting with agentic workflows who need visibility into what each agent is doing and the ability to approve or redirect at any step.

Product and operations teams use Mercury to coordinate multiple AI agents on one canvas and delegate tasks with human approval.