Describe the job
Tell Works what you need. AI drafts the employee’s role and instructions, with a live preview as it builds.
Describe the employee you need. Give it the context to do good work. Keep the whole team in view.
Works is the AI employee management system we’re building: a place to create roles, assign tasks, follow progress, and handle the decisions that need you.
Build your next employee
An employee who turns my meeting notes into clear action plans.
Operations assistant
Document analyst
From a job description to a working role
Tell Works what you need. AI drafts the employee’s role and instructions, with a live preview as it builds.
Review the draft, refine the instructions, and add relevant workspace files. You decide what good work looks like.
Assign a task, follow the run, and review the result. Keep the employee’s context and history together for the next job.
AI employees, connected CLI sessions, and Relay window automation share one workspace.
Describe the work you want off your plate. Works uses your connected model to draft a name, role, and instructions. Follow the draft as it’s generated, then edit it before creating your employee.
Add notes, instructions, and text files to an employee’s workspace. Keep the information it needs alongside the job it’s meant to do.
Give employees individual tasks or configure recurring work. See recorded actions, results, and run history. Pause employees and review proposed file changes before they’re saved.
Connect supported CLI sessions to bring their activity into view. Works’ manager collects progress, surfaces blockers, and prepares briefings so you can see where your attention is needed.
Watch employee creation stream into view and follow task activity as it’s recorded. Know what’s running, what finished, and what needs a decision.
Browse employees, refine their roles, inspect their files, and review their work. A shared management layer for your AI employees and connected workers.
Introducing Relay · formerly Worker
Open Relay in Works to attach a browser or desktop app window. Give it a task, or use Watch and learn to demonstrate a workflow. Review the steps, save a routine, and follow its next run in the live activity view.
Search your open windows, copy task logs, and delete or re-record saved routines. Relay uses the native Mac app and its existing Screen Recording and Accessibility permissions.
Explore RelayWhat we’re building today
Works is currently a local application in development. Employee configurations, workspace files, and run history live on your machine. Tasks and schedules run while the Works runtime is running and the machine is awake.
You connect a compatible model for employee creation and tasks. If you choose a cloud provider, it receives the prompts and workspace content used for that work. Connected CLI workers use their own model connections and permissions.
Relay brings window automation into Works. Its native Mac app handles screenshots, desktop input, and saved routines. You can run tasks and teach workflows from the Relay section; this does not automatically grant desktop control to your AI employees.
We’re building toward a simpler way to direct a team of AI workers, with useful progress reports and a clear place for human decisions.
Works is in development. Tell us about the workflow you’d like to manage.