LearnGrokNothing here is clinical advice, and nothing here should touch identifiable patient information.
You need a weekly plan that a service manager can review before anyone changes appointment capacity. This pack turns aggregate demand, rota gaps, rooms and service rules into a draft timetable, an exception list and clear operational decisions. It is for clinic operations managers and scheduling leads, not for clinical prioritisation.
Keep every input free of identifiable patient information. Use counts by service, appointment type, day and time band. Do not paste names, dates of birth, record numbers, free-text referrals or any other patient-level data.
Key point
Build the evidence trail first
A useful capacity plan shows which source or assumption supports every proposed slot and every shortage.
Prepare the four planning documents
Before using the prompts, create four working documents for the same week.
- Export an aggregate demand extract. Include service, appointment type, slot duration, booked count, waiting count and demand by day or time band where available.
- Export the staff rota. Use role labels or internal non-personal labels. Include shifts, breaks, protected time, permitted services and unavailable periods.
- Export room availability. Show opening periods, closures, equipment restrictions and service suitability for each room.
- Copy the current service rules into one document. Include session lengths, required roles, setup time, booking caps and any approved booking restrictions.
Use the first prompt to check whether those documents can be reconciled. It does not build a timetable. It identifies the errors that would otherwise create a convincing but impossible schedule.
Watch out
Do not turn incomplete data into false certainty
If a rota does not state whether a role can cover a service, leave the coverage unresolved. Do not assume that a staff member or room is interchangeable.
There is no universal rule for how much material the model can handle at once. This is version-dependent, so check the xAI documentation overview if you need to divide a large weekly export. When splitting material, keep a stable service label, appointment-type label and planning-week date in every part.
Build the schedule in two passes
Run Check the planning inputs first. Correct source data where you can. Where a correction needs a manager decision, retain it as a provisional assumption. Copy the resulting assumptions into the next prompt.
Then run Build the proposed weekly schedule. The schedule table is deliberately detailed. You need to see the date, time, room, role coverage and slot count on the same row. A summary with only total appointments is not enough to identify a room clash or a missing staff role.
Do not paste an existing patient booking list. The purpose is to plan capacity, not to select, move or contact individual patients. If your local process requires booking order, provide only the approved operational rule, such as a stated booking window or referral queue category. The qualified service team decides any clinical matter.
Check
Check each proposed session against the source files
Pick several rows across the week. Confirm the room is open, every required role is on rota, breaks are protected and the slot count follows the stated rule.
Separate shortages from causes
Run Find unmet demand and constraint failures after the draft schedule. This prompt reconciles demand with allocated slots and makes unallocated but feasible capacity visible. That distinction matters. A shortage caused by no suitable room needs a different decision from a shortage caused by a missing rota role.
Treat the reconciliation as a control. For each service and appointment type, demand should equal allocated demand plus unmet demand. If it does not, the output is wrong or the input definitions do not match. For example, one extract may count requested appointments while another counts only bookings within the planning week.
Use this table when reviewing common results:
| If you see | Check first | Next action |
|---|---|---|
| Unmet demand with unused slots | Service, role and booking restrictions | Confirm whether the unused slots are genuinely eligible, then amend the draft only if rules permit |
| A staff conflict | Rota times, breaks and permitted services | Correct the rota source or mark the session contingent |
| A room conflict | Room closure, equipment and service rules | Move the session or identify the room constraint for decision |
| A reconciliation mismatch | Demand definition and appointment-type labels | Re-run after aligning the source categories |
Put choices in front of the manager
Use Prepare service manager decisions only after the audit. It compares operational options without pretending that every option is available. An extra session might require a room, an approved role and a permitted staffing arrangement. If any one is unknown, the option remains contingent.
The decision table should make three things clear: capacity gained, effect on another service, and approval needed. A recommendation can be operationally sound while still awaiting workforce, facilities or service-manager approval.
Note
A booking restriction is a managed choice
Record its wording, scope, trigger and approval. Do not present a restriction as an automatic response to unmet demand.
Finish with Create the review pack. Send the pack as a proposed plan. Keep the original rota and room files beside it, so reviewers can trace each change. After approval, update the live scheduling system through your local process and retain the change log.
When the output does not work
Stop if the output invents a room, staff capability, operating hour or booking rule. Remove the unsupported row, return to the input-check prompt and supply the missing source or ask the named owner to decide. If tables are incomplete, split the week by service or by day, then combine only results that use the same assumptions. If unmet demand cannot be reconciled, do not publish the plan as ready to book. Mark it as unresolved and take the mismatch to the service manager.
