SMF·WHEN-I-WORK
Can a prompt replace When I Work?
Legal, HR & payroll — employee scheduling
Exhibit tracking slip
Verdict
Two dollars fifty a head is priced against the effort of not doing it, and the software underneath is a weekend. The gap is adoption: this only works if every member of staff has it on their phone and opens it, and an installed app with push notifications is a different thing from a web page they have to remember. Nothing here is a network or a dataset, so kinda is right.
Exhibit A — The prompt
Received on31.07.2026Build a scheduling tool whose distinguishing feature is that shift changes resolve without you.
Stack: your choice, Postgres, Docker Compose, a domain with TLS. Mobile-first, keyboard-free, usable on a slow connection.
Schedule: employees with roles and stated availability, shifts with role, start, end and required headcount, published a week at a time.
Availability: each employee sets recurring availability and one-off exceptions. Assigning someone outside their availability is allowed but warns and records that it was overridden — a hard block just gets worked around by text message.
The self-service loop, which is the point:
- **Offer a shift up**: an employee releases a shift into an open pool.
- **Claim from the pool**: anyone with the right role and matching availability sees it and can claim. Claiming is atomic — two people tapping at the same second must not both get it, and this needs a real transaction, not a check-then-write.
- **Direct swap**: propose a trade with a named colleague, who accepts or declines.
- Every resolution needs one manager approval and nothing else. The manager sees a single queue of pending changes with accept and reject.
- Rules enforced before a claim is offered: not already working that time, not exceeding a weekly hours cap, and a minimum rest gap between shifts.
Notifications: web push where available, email otherwise, for published schedules, changes to your own shifts, and a shift available in the pool you qualify for. Be honest in the README that without a native app this is where a self-hosted version is weakest.
Show, on the schedule, the coverage per role per hour as a small bar, so an understaffed Saturday is visible while it is still a draft.
Write tests for atomic claiming under concurrency, for the rest-gap and weekly-cap rules blocking a claim, and for availability overrides being recorded.
Do not build a time clock, payroll, or multi-location support.Opening prefills the prompt — press enter to run it.
Exhibit B — What you lose
- B.1 the mobile apps and push notifications
- B.2 the time clock and its payroll export on the paid tiers
- B.3 multi-location scheduling
- B.4 availability requests handled entirely in-app by staff
- B.5 somebody else keeping it up on a Friday night
Exhibit C — Why people still pay: mobile adoption by hourly staff
Because the whole value is that the rota reaches everyone and changes reach them too, and that requires an app people already have rather than a link you keep resending.
Questions
Why does atomic claiming matter?
Because two people claiming the same shift is a guaranteed argument, and a naive check-then-write gets it wrong exactly when the pool is busy. One transaction with a conditional update solves it and needs a test to stay solved.
Why warn rather than block on availability?
Because managers legitimately need to ask someone to work outside their stated availability, and a hard block pushes that conversation into text messages where the schedule cannot see it. Warning and recording keeps it in the system.
How much does the lack of an app really cost?
A lot, honestly. Hourly staff will not check a web page, and web push is inconsistent on iOS. Expect to also send an SMS for anything urgent, which is a real cost the subscription absorbs.
What does When I Work cost?
$2.50 per user per month for a single location, $5 for multi-location, $8 for the top tier — with the time clock and payroll export on the higher tiers rather than the entry one.
Related tools
Receipt