When a provider calls out, the assistant proposes the fix

The schedule you printed at 7 a.m. is not the schedule the team actually ran.

By 8:15 a PT has called out for the day. By 10:30 a counselor is leaving early for an emergency. By lunch a Speech evaluation has been pulled forward at the request of discharge planning. Every one of those events ripples — the patients who were on the missing provider’s list need to land somewhere else, and “somewhere else” has to keep faith with every preference and restriction the facility has on file.

Most of the day’s scheduling work is not the Sunday-night build. It is the dozens of small recompositions that happen between Monday morning and Friday afternoon.

What goes wrong when someone calls out

The standard workflow, in most facilities, is this. The scheduler gets the call. They open the day’s grid, scroll to the absent provider’s column, and start reading down it. For each appointment, they ask:

• Who else can cover this discipline this hour?
• Is that person already busy?
• Does the patient have a preferred provider on shift?
• Is there an avoid or required relationship in play?
• Did this patient already see this substitute today, and is that fine or is it the wrong vibe?
• Am I about to put two patients with a specific-patient restriction into the same group because the substitute happens to lead that group?

This is the kind of work that takes thirty to ninety minutes per absence and still gets it wrong some percentage of the time, because the constraints live in the same place they always have — partly on the screen, partly in the scheduler’s head. The bigger the unit and the more rules in play, the worse the hand-rebuild scales.

What Tempo’s assistant does instead

When the scheduler tells Tempo a provider is out — full day, half day, single block, late arrival, early leave — the assistant runs the same composition pass the engine uses on Sunday night, but scoped to the affected window only.

For every appointment that was on the absent provider’s calendar, it proposes a path forward:

• Reassign the appointment to a substitute who is available, in the same discipline, and not on this patient’s avoid list. If the patient has a preferred provider on shift, prefer that one.
• If a required-provider relationship would be violated, surface the appointment as a gap rather than silently breaking it.
• If the absent provider was leading a group, recompose the group’s provider mix and re-check every patient roster against the new lead’s exclusions. Anyone who no longer fits is moved out of the group and rescheduled for a make-up slot.
• If no clean reassignment exists, propose the best partial fix and flag the unfilled minutes against the patient’s weekly target.

The assistant produces a diff: this is the schedule you had, here is the schedule it is proposing. Each changed cell is a discrete proposal.

Approve or reject, one chip at a time

The piece that matters as much as the proposal itself is the approval surface.

A wholesale “rebuild the day” button would either ask the scheduler to trust the assistant blindly or force them to audit every cell after the fact — both bad. Instead, every proposed change in the diff is a separate approve / reject:

• Reassignment of a 1:1 appointment to a different provider — approve to commit it, reject to leave the appointment unassigned and handle it yourself.
• Removal of a patient from a group whose new provider lead is on their avoid list — approve to pull them and let the assistant find a make-up slot, reject to keep them in the group and override the exclusion manually.
• Time shift of an appointment to a different block — approve, edit the time, or reject and keep the original.
• A flagged gap with no proposed fix — acknowledge it so it shows up on the day’s exception report instead of disappearing.

Nothing commits until the scheduler approves it. Reject any individual proposal and the affected cell snaps back to its previous state. Approve them all and the day’s grid updates in a single operation, preferences and restrictions intact, and the affected patients, providers, and family portals see the new times without anyone having to redistribute a PDF.

Why this is the part that saves the day

We have written before about the Sunday-night build and about the rules the engine is reading from. Those are real problems and Tempo solves them. But the part of the job that actually fills the scheduler’s calendar is the in-week churn: the call-outs, the early leaves, the pulled-forward evaluations, the patient who refuses the substitute and wants to wait for their usual person.

Asking the scheduler to rebuild the day from scratch every time the day moves is the workflow that turns the role into a fire department. Asking them to approve or reject one proposal at a time — proposals that already respect every restriction on file — turns it back into oversight.

If your week loses an hour every time a provider calls in sick, join the Tempo waitlist and we will show you what it looks like when the assistant proposes the fix and you just say yes or no.