Scheduled Batch Screening
Batch Screening can run on its own, at whatever time you set. The Schedule button, next to the start button, saves the screening you just configured — same context, same mode — for a specific date or for a recurrence.
The Datta Scheduler fires it, the same clock that runs DATTA Extract loads. It is a scheduler belonging to the platform itself, with the same guarantees as the loads: one execution per slot, a record of what was missed while the platform was down, and the same calendar.
How to schedule
- Open Process › Batch Screening.
- Pick the case context (the coloured chip at the top).
- Pick the screening mode — this is what travels with the schedule:
| Mode | The schedule will repeat |
|---|---|
| Screen new documents | screening what came in and has no report yet |
| Reprocess errors | re-running only the rules left without analysis |
| Reprocess all (including already screened) | deleting the reports and screening from scratch |
| New/updated rules only | re-evaluating only rules created or changed after the last screening |
- Click Schedule.
- In the dialog, pick the recurrence:
| Recurrence | When to use it |
|---|---|
| Once | a specific date and time. Runs once and the schedule ends by itself |
| Every | free interval — every 5 minutes, every 90 minutes, every 2 days |
| Daily | one or more times of day, every day |
| Weekly | chosen weekdays, at the given times |
| Monthly | chosen days of the month. A day that does not exist in a month (31 in February) is skipped |
- Check the next executions the dialog lists — they are computed by the server, with the same maths the scheduler will use. What you see is what will run.
- Save schedule.
The schedule appears right below the buttons, under Scheduled screenings, with the recurrence in plain language, the context, the mode and the next run.
The mode travels with it — and it is the same one the start button uses
The schedule stores exactly what the Start Screening button would have sent at that moment. If you want two different routines — say, "screen new" every night and "reprocess errors" on Sundays — create two schedules: pick the mode, schedule it, change the mode, schedule again.
To change the mode of an existing schedule, use Edit on its card: the dialog opens with the current recurrence and saves whichever mode is selected on the screen at that moment.
Pause, resume, edit and remove
Each schedule has four actions:
| Action | What it does |
|---|---|
| Pause | keeps the schedule and stops firing. Nothing is deleted |
| Resume | puts it back on the clock; the next run counts again |
| Edit | reopens the dialog to change the recurrence (and the mode selected on screen) |
| Remove | drops the schedule for good. Reports already written are preserved |
Pausing takes effect immediately and on every replica: what decides the firing is the stored state, re-read at each occurrence.
What happens at the scheduled time
The scheduled screening starts exactly as if you had clicked Start: reports are written case by case and you follow the progress by opening the screen.
Two outcomes deserve an explanation:
- A screening was already running for that context. The platform does not start a second one — the one in flight continues, and the scheduled occurrence is recorded as "nothing to do". This is not an error and does not stop the schedule.
- The platform was down at that time. A missed execution is always recorded. What happens next is what you chose in the dialog: "Skip the execution" (default) or "Run once when back" — which replays one execution, not one per missed slot.
Where else the scheduled screening shows up
It is a line on the platform's agenda, so it also appears under DATTA Extract › Scheduler:
- Calendar — alongside the loads, on the predicted day and time, labelled Batch screening. The card's button brings you back to this screen.
- Engine — in the list of what the clock will fire, with the next run, the last fire and the missed executions. Whoever administers the platform can pause it there.
Permissions
| Action | Permission |
|---|---|
| See the scheduled screenings | AUDIT_VIEW |
| Create, edit, pause, resume and remove | AUDIT_SCHEDULE |
| Pause/resume from the Scheduler's Engine tab | PLATFORM_ADMIN |
The default analyst and advanced user roles already get AUDIT_VIEW and AUDIT_SCHEDULE — whoever screens also schedules. The permission is separate from AUDIT_EXECUTE (firing the screening) on purpose: firing is an observed act, with someone watching the screen; scheduling leaves the platform screening on its own indefinitely. An organisation that wants to keep manual screening and remove the automatic one can revoke AUDIT_SCHEDULE alone.
Without the permission the Schedule button does not appear — and a direct call is refused with 403 and a message in Portuguese.
FAQ
Does the schedule disappear after it runs? Only the Once type: it fires on the chosen date and ends by itself, with nobody having to remember to switch it off. The other recurrences keep going.
Can I schedule a date in the past? No. The schedule is refused right away, with the message "Este agendamento não tem nenhuma execução futura. Revise a data ou a janela." — instead of being saved as something active that never fires.
Can I schedule the screening of one specific case? Not here. The schedule is for batch screening, over a whole context.
Does a scheduled screening consume AI quota? Yes, exactly like the manual one. That is why the mode matters: "Reprocess all" redoes everything from scratch on every run and is the most expensive; "Screen new documents" only evaluates what has no report yet.
See also
- Batch Screening — the manual run, the modes and the report.
- Load Schedule — the Scheduler calendar.
- Scheduler — Engine tab — operating the scheduler, for platform administrators.