Scenarios · Practice operations
Coverage rosters that are computed, not negotiated.
A practice manager declares the staffing rules once — every shift needs a qualified nurse, nobody works a day shift straight after a night, weekend load stays fair — and the roster is computed from them. Not suggested by an AI that might miss a rule: computed, so the schedule either satisfies every rule or names the exact rules that collide. The manager reviews and publishes; the run lands on the audit trail like everything else in HASP.
What changes for your team: publishing next month's roster becomes a review step, not a negotiation — and any challenged schedule can be replayed from the rules that produced it.
Where the work happens
- Studio — staff, shifts, and roster requests live as plain records; the scheduling workflow and its rules are edited like any other workflow.
- Assistant — describe a rule change in plain language ("no more than three consecutive days for anyone") and Studio reformulates the schedule's rules for your review.
- Workflows — a solving step computes the assignment; downstream steps record the result and notify the team.
How it works
-
The rules are declared, not buried in a spreadsheet.
Coverage minimums, qualifications, availability, rest requirements, and fairness preferences are stored as a structured rule set on the workflow — readable in plain language, versioned like everything else, and editable without touching a formula.
-
A roster request fires the scheduling workflow.
The workflow's solving step evaluates the rules with a deterministic scheduling engine — the same inputs always produce the same roster, and the computation runs entirely inside your HASP environment. No AI model touches the assignment math.
-
A valid roster comes back — or the conflict does.
When the rules can be satisfied, every assignment provably respects all of them. When they can't, the workflow reports exactly which rules collide — "you require an RN on both Tuesday shifts, but only one RN is available Tuesday" — so the tradeoff is a decision, not a mystery.
-
A human publishes the schedule.
The computed roster is written back as a draft for the practice manager to review and release. Who published which roster, computed from which rules, lands on the audit trail before the schedule reaches the team.
Try this as a Studio template.
Why this survives governance
-
Schedules encode legal obligations — guessing at them isn't an option.
Rest periods, qualification coverage, and hour caps come from labor law, licensure rules, and union agreements. A language model drafting a schedule can silently violate any of them and look fine doing it. A computed roster can't: every rule you declared is either satisfied or reported as unsatisfiable — there is no third state.
-
The same rules always produce the same schedule.
Because the computation is deterministic and its inputs are recorded, any published roster can be reproduced after the fact from the rules and roster request on the audit trail. "Why did Jordan get three weekend shifts?" has a checkable answer, not a shrug.
-
Fairness disputes get evidence instead of memory.
The fairness preference is a declared, weighted rule — not a manager's recollection of who worked last Christmas. When the schedule is challenged, the record shows the rules in force, the request, and the computed result, end to end.
Deploy this workflow in your environment.