The work nobody schedules, on a schedule.

Put recurring work on a timer: the postponed refactor, the drifting documentation, the slowly slowing tests. Answer agents from Slack, hear from them by email, and connect the systems you already run.

Plugs into everything. Politely.

Answer from Slack. The run waits for you.

A guided task posts its question to you in Slack. Press a button or reply in the thread, and the same run carries on. The console shows the same question, so whoever sees it first answers, and notifications reach you by email too.

  • The question, posted to the person who filed the task
  • A button press or a reply is the answer
  • Notifications in Slack or by email
Two questions an agent asked before deciding, each with three options and the answer the owner chose.

Recurring work, filed on time.

Jaah. It keeps working while you are away. A schedule files a real task at each firing — instructions, project, persona, model — and that task runs like any other: its own copy of the code, your tests, the reviews, a merged change.

  • A cron expression, or a fixed interval
  • Instructions, projects, persona and model per schedule
  • Every firing listed, with the task it filed
  • A schedule that is overdue or stuck is flagged

What keeps slipping? Put it on a timer.

Sign up and pick the plan that fits what keeps slipping in your repositories, and how often you would want it done.