How Do I Schedule Recurring Tasks With a Desktop AI Assistant?

Jarvis (getjarvis.eu), the screen-aware desktop AI assistant for macOS 12+, Windows 10+, and Linux, lets you schedule recurring tasks in plain language from its floating bar (Cmd+/ on macOS, Ctrl+/ on Windows). Open the bar and say something like "every weekday at 8am, summarize my unread Gmail and post it to my Notion daily log." Jarvis stores the schedule, then runs the multi-step job on its own, routing each step to the best model (frontier models from Anthropic, OpenAI, or Google) and using your connected apps. You can list, edit, pause, or delete schedules anytime. It works across 30+ connectors including Gmail, Slack, Notion, Linear, Google Calendar, and Todoist. Plans start with a free plan (40 requests/week), then Pro at $16/month. Tasks run while the app is open and online; a closed laptop can't fire a schedule, so Jarvis is honest about that limit.

You never write cron syntax. Summon the floating bar with Cmd+/ (macOS) or Ctrl+/ (Windows) and describe the cadence conversationally: "every Monday at 9am," "each weekday morning," or "on the 1st of the month." Jarvis parses the timing, confirms the exact recurrence back to you, and saves it. Because Jarvis is screen-aware, you can also reference what's in front of you, e.g. "schedule a weekly export of this Airtable view." Each saved task keeps the full instruction, so a single schedule can span several apps and steps rather than one isolated action, and you approve it before it goes live.

Ask Jarvis to "list my scheduled tasks" and it shows every active recurrence with its next run time. You can pause one for a busy week, change the time, swap the target app, or delete it entirely, all in conversation. This matters because automation drifts: the morning brief you wanted in January may need a different format by March. Keeping schedules editable from the same floating bar, rather than buried in a separate dashboard, keeps you in control. Jarvis routes the underlying work to frontier models from Anthropic, OpenAI, or Google depending on the step, so a summarization task and a data-extraction task each use a suitable model.

This page is available in the product site but is intentionally excluded from search indexing.

Capabilities