Scheduling
The scheduling tool works out who does what, and when, under rules you give it: shift rosters, job assignments, league fixtures, room bookings.
Who can see this: anyone who can open the project. Solving a schedule adds steps, so you need to be able to edit the project.
Read the transcript
What the video shows. It has no narration, so this is a description rather than a transcript.
Seven nurses, twenty-one shifts and forty-seven rules, laid out as a week of morning, afternoon and night slots, with the question “Can you solve this by hand?”
Step 1, Describe your rules. The rules you typed are sorted into hard constraints, each one a line: every shift needs exactly one nurse, no back-to-back shifts, no morning after a night, and each nurse’s weekly maximum respected. Underneath, the path from what you wrote to what gets solved: prompt, rule extraction, hard and soft classification, then the model.
Step 2, Solve it. The empty roster fills in, and the result reads twenty-one shifts covered, every constraint passed, and OPTIMAL — solved in under half a second.
- Result: whether the schedule is the best possible, or just a good one.
- Rules Applied: each rule, and whether the schedule meets it.
- The schedule: who or what is assigned where.
When to use it
Section titled “When to use it”Use it when you have things to assign and rules to follow:
- staff shifts with coverage and rest rules
- jobs across machines or people, with time limits
- fixtures, bookings or deliveries across slots
Ask for it
Section titled “Ask for it”Ask in project chat, with the people or resources, the things to assign, and the rules. For example:
Assign these shipments to 3 packers, max 8 hours each.
Your rules can come from two places:
- A rules table. If one of your tables lists the rules, say which one. Querri reads the rules from it.
- Your request. Otherwise Querri writes out the rules from your request and your data.
Either way, each rule is either hard (must never be broken) or soft (a preference to aim for). Keep the data and rules as separate tables; don’t join them first.
What runs
Section titled “What runs”- The Scheduling step reads the rules, then uses a solver to find the best schedule it can — Google’s OR-Tools CP-SAT, which is what the video names on screen. Each attempt gets 60 seconds.
- A chart step draws the schedule.
What you get
Section titled “What you get”The step’s findings come in four parts:
| Part | What it tells you |
|---|---|
| Result | OPTIMAL (proven best) or FEASIBLE (a good solution, may not be the best), and how long it took |
| Rules Applied | Each rule in plain words, marked PASS or FAIL |
| Key Numbers | The few figures that matter, like shifts per person or total time |
| What This Means | A short explanation anyone can follow |
It also adds a table of the schedule to the project.
When no schedule fits
Section titled “When no schedule fits”If the hard rules contradict each other, there’s no schedule that meets them all. The step stops with “Could not find a feasible schedule.” and suggests which rules to relax. Loosen one, or turn a hard rule into a soft one, and ask again.
Refine a schedule
Section titled “Refine a schedule”To change one thing, like a rule or a preference, use Refine on the result instead of starting over. The solver’s code is split into labeled sections for data, rules, objective and results, so a small change stays small.
Refining changes only that step. Run the project again to update the steps after it, such as the chart.
Credits
Section titled “Credits”Asking uses a credit for the question, as usual. The scheduling tool doesn’t add a separate charge. Your plan may differ; see How credits are counted on Monitor usage.
- Say what “best” means. Balanced workload, fewest late jobs, least travel: name the goal.
- Mark must-haves as hard, and leave nice-to-haves soft so the solver has room.
- Try one week or one team first, check Rules Applied, then scale up.