We’re building Ever Works in public. Some things may be incomplete, missing, or broken while we continue improving the platform.We’re building Ever Works in public — expect a few rough edges.
Feature · Schedules
An AI team does not keep office hours, but it should keep yours. In Ever Works nothing runs on a clock until you arm one: a Work refreshes on the cadence you pick, a Mission ticks on a cron you write, an Agent wakes on its own heartbeat, and a Task can recur. Clocking in is setting that schedule; clocking out is pausing it. Every timer in the account then shows up in one Schedules view, with what owns it and when it runs next.
Each schedule belongs to the thing it drives, so you set it where you already work.
On a Work’s Worker tab, open Schedule and pick one of seven cadences: hourly, every 3, 8 or 12 hours, daily, weekly or monthly. A failed run is retried after 15 minutes, and after three consecutive failures (configurable from one to ten) the schedule pauses itself instead of burning spend. Run now skips the wait.
A Mission starts as one-shot. Flip it to scheduled and give it a five-field cron, where 0 9 * * * means every day at 09:00 UTC. Each tick proposes new Ideas toward the goal and holds off once the outstanding-Ideas cap is reached, which defaults to 20.
Give an Agent a cron cadence or leave it on manual. On each tick it is asked for the single most useful next action: create a task, comment on one it is part of, record a learning in its own instructions, or just observe. You can tune that per Agent to propose, observe or do nothing.
A Task’s Schedule panel can make it recur (daily, weekly, monthly or a custom rule, with an optional end date and a cap of up to 9,999 occurrences) or arm it for one exact moment.
Not every shift starts on a timer. An inbound trigger gives you a signed webhook URL, and every verified call spawns a Task for the Agent you chose, started straight away unless you asked it to wait in the backlog.
Stopping work is as deliberate as starting it, and nothing you pause is thrown away.
Pause and Resume sit in every Agent’s header, next to Run heartbeat now. An Agent that fails three runs in a row pauses itself; you can set that anywhere from one to twenty.
The tick worker skips a paused Mission, and the Ideas and Works it already produced stay as they are. Resume it and it ticks on its cron again.
A paused schedule keeps its cadence and settings until you switch it back on. Cancelling it is the stronger step: the cadence is cleared and the schedule returns to its default state.
From Sessions, interrupt a running Agent and it stops cleanly, keeping the work it has already committed. Resume it later and it continues from where it stopped.
Open Activity and switch to Schedules. Recurring Tasks, Agent heartbeats, Work schedules, Mission ticks, source validation, data sync and inbound triggers are listed together, sorted by next run, and each row shows its Owner, Cadence, Next run and Status. The dashboard’s Coming up block shows the next three scheduled Work or Mission runs. The Schedules view is read-only on purpose: follow the Owner link and change the cadence where it lives, so it is never edited from two places at once. A few switches sit underneath all of this, and they are worth knowing before you plan a routine around them.