AI Shift Scheduling vs a Rota Spreadsheet for Small Teams

For a team under about eight people with a stable weekly pattern, a rota spreadsheet with built-in checks, drafted by an AI assistant, is usually enough. Past roughly ten staff, or with changing availability, frequent swaps or strict hours rules, an app with auto-scheduling (Deputy from $6.50 a user a month, When I Work from $2.50) pays off.

The real difference isn't the AI. A scheduling app collects availability on people's phones, handles swaps without the manager, and enforces rules every time. A spreadsheet depends on one person remembering everything. And general AI assistants, asked to build a rota from scratch, are surprisingly bad at hard rules: they'll happily give someone two shifts that overlap.

Follow me on Instagram@sagnikteaches

What each option actually does

Write the rota rules down before comparing anything

Whether you stay with a spreadsheet or move to an app, the rules have to exist in writing, and most small teams have never written them. The list is usually short. The kitchen fitter's showroom ended up with nine lines:

Connect on LinkedInSagnik Bhattacharya

• Opening hours: weekdays 09:00-17:30, weekends 10:00-16:00.
• Minimum two people on the floor during opening hours; three on Saturdays between 11:00 and 14:00.
• At least one designer on the floor at all times.
• Designers: five days a week, two consecutive days off, alternating weekends.
• Each part-timer's contracted maximum hours, and any days they can't work.
• No more than six days in a row for anyone.
• At least 11 hours between one shift ending and the next starting.
• Holidays booked at least four weeks ahead are protected; later requests depend on cover.
• The rota is published by Wednesday for the following week.

Writing those nine lines took 20 minutes and exposed a rule that lived only in the owner's head: the Saturday lunchtime third person. Every missing rule is a mistake waiting to happen, whether a person, an AI or an app draws up the rota.

Subscribe on YouTube@codingliquids

What happened when an AI assistant drafted a rota

To see where general AI helps and where it doesn't, take an illustrative kitchen fitter's showroom. It opens seven days a week (9-5:30 weekdays, 10-4 at weekends) and needs two people on the floor at all times, one of whom must be a designer. There are six part-time staff and two full-time designers. A prompt of the kind an owner might write:

Build a weekly rota for our kitchen showroom as a table (rows = people,
columns = days, cells = shift times or OFF).

Opening hours: Mon-Fri 09:00-17:30, Sat-Sun 10:00-16:00.
Always 2 people on the floor; at least 1 must be a designer.
Designers (full time, 37.5 hrs, 5 days, 2 days off in a row):
  Designer A, Designer B.
Sales staff (part time): S1 max 24 hrs; S2 max 20 hrs, never Sundays;
  S3 max 16 hrs, weekends only; S4 max 24 hrs; S5 max 20 hrs, not Mondays;
  S6 max 12 hrs.
Nobody works more than 6 days in a row. Minimum 11 hours between shifts.
After the table, list each person's total hours.

The draft came back in seconds, neatly formatted, and looked plausible. Checking it cell by cell found four problems:

• Saturday afternoon had two sales staff and no designer, because Designer B's two days off had been placed on Saturday and Sunday while Designer A was also off on Saturday.
• S2 was rostered on a Sunday despite "never Sundays".
• The hours total said S1 worked 24 hours; the cells added up to 26.5.
• Tuesday had three people on the floor all day, which wasn't wrong but wasted about eight paid hours.

That's typical. Language models are good at producing something rota-shaped and poor at satisfying many interlocking rules at once, and they're unreliable at adding up their own totals. Asking a second time ("check this rota against every rule and list every breach") caught three of the four. The lesson isn't "don't use AI". It's that the AI draft needs a mechanical check, from formulas or from software built for scheduling.

Building the spreadsheet so it checks itself

A rota spreadsheet becomes far more reliable with a few checking formulas, and an AI assistant is good at writing those for you. A sensible layout: one row per person per day, with columns for date, person, role, start and end. From that "long" table, a separate tab can show the familiar grid, and checks run on the long table. Useful checks:

Hours per person per week (in a summary tab):
=SUMPRODUCT((Rota[Person]=A2)*(Rota[Week]=B2)*(Rota[End]-Rota[Start]))*24

Designer on the floor for each opening slot (1 = yes):
=IF(COUNTIFS(Rota[Date],D2,Rota[Role],"Designer",
   Rota[Start],"<="&E2,Rota[End],">="&F2)>=1,1,0)

Rest gap before a shift, in hours (flag if under 11):
=(this shift's start - that person's previous end)*24

Add conditional formatting so any breach turns red, and a single "problems" cell at the top that counts the red cells. The rota is ready to publish only when it reads zero. If you're working in Google Sheets, prompts for Google Sheets AI covers getting Gemini to write formulas like these; in Excel, Copilot can do the same on a licensed plan, and Microsoft 365 Copilot Business is $21 a user a month on annual billing.

Time to build: two to three hours the first time, most of it defining the rules. After that, the weekly rota takes the manager perhaps 45 minutes instead of two hours, because the sheet catches mistakes rather than the staff.

Collecting availability without buying an app

The most time-consuming part of a spreadsheet rota is often chasing who can work when. A simple weekly form fixes most of it without new software. Create a short form in Microsoft Forms or Google Forms with three questions (name from a dropdown, days unavailable next week, and any shift preferences), set responses to flow into a tab of the rota workbook, and send the link every Monday with a Wednesday-noon deadline.

An AI assistant is useful at the next step. Paste the week's responses and the rules, and ask it to list clashes rather than to build the rota: "Which days next week can't meet the minimum staffing or designer rule, given this availability?" A typical answer, illustrative for the showroom: "Thursday has only one designer available (Designer A is on holiday), and S2 and S5 are both unavailable on Saturday, leaving three possible sales staff for a day needing three people at lunchtime." That's the kind of question AI handles well: spotting problems in a small amount of data, which you then solve with a phone call. It avoids asking it to satisfy every rule at once, which is where it stumbles.

If swaps, rather than availability, cause most of your pain, a form won't help much. That's usually the signal that a free app tier is worth trying.

The scheduling apps a small team would look at

List prices checked in September 2026; confirm on each vendor's page, because features move between tiers.

Two practical notes. Per-user pricing looks tiny until you count everyone who needs a login, including occasional staff. And minimum spends matter at small sizes: at Deputy's $30 minimum, a team of four on Core pays $30, not $26.

Choosing between them

The kitchen fitter's decision, costed

The showroom's eight people (six part-time sales staff and two designers) plus two fitting crews of two. The crews' work is job scheduling rather than shifts: they're allocated to installations, which lives in the job-management software and is a different problem, covered in AI job scheduling and dispatch for field teams. So the rota question is really about the showroom's eight.

Before, the owner spent about two and a half hours every Sunday on the rota, and roughly an hour a week afterwards dealing with swap requests in a group chat. Mistakes surfaced as a Saturday with no designer, around once a month, which meant the owner coming in.

Option one was the self-checking spreadsheet. Cost: nothing extra, since they already had Microsoft 365. Estimated time: 45 minutes a week drafting plus the same hour on swaps. Option two was Deputy Core for eight users: 8 × $6.50 = $52 a month. Estimated time: 20 minutes a week reviewing the auto-scheduled draft, with swaps handled in the app and only approvals left for the owner.

The difference came to about an hour and a half a week, around six hours a month, for $52. The owner chose the app, mainly for the swaps: the group chat had become the most irritating part of the job. After a month, the Saturday-without-a-designer problem had stopped, because the designer requirement was a rule the app enforced rather than something to remember.

For a showroom with three staff and fixed hours, the same sums would have gone the other way. There's no prize for using an app you don't need.

A locksmith's on-call rota: a spreadsheet job

A two-van locksmith with four people sharing night and weekend call-outs needs a rota, but not a scheduling app. The pattern is simple (one person on call each night, one each weekend day), and the real issue is fairness: who has done the most nights, weekends and public holidays over the year.

A spreadsheet with a running tally does it, and an AI assistant helps with the awkward part: proposing next quarter's rota so the totals even out, given everyone's booked holidays. A prompt and an illustrative output:

Prompt: Here are this year's on-call counts so far (nights / weekend days /
public holidays): P1 41/12/2, P2 38/14/1, P3 44/10/2, P4 35/12/0.
Holidays next quarter: P1 week 3, P4 weeks 8-9. Propose an on-call pattern
for the next 13 weeks that brings totals closer together. Nobody does more
than 2 nights in a row. Show the updated totals.

Output (extract): Week 1: Mon P4, Tue P4, Wed P2, Thu P1, Fri P2,
Sat P4, Sun P3 ... Projected year totals: P1 60/17/2, P2 61/18/2,
P3 60/17/2, P4 59/18/1.

What you'd check: that nobody is rostered during their holiday weeks, that no one has three nights running, and that the projected totals really follow from the pattern. Recount them in the sheet; the AI's arithmetic is the part most likely to be wrong. Used this way, as a proposal you verify, AI saves most of the head-scratching without being trusted with the final answer.

An HVAC installer's mixed rota

Some teams need two kinds of rota at once. An illustrative ten-person HVAC installer has an office open 8 to 5 on weekdays with three staff, six engineers whose days are filled by jobs rather than shifts, and an out-of-hours on-call rota shared by four of those engineers during the winter months.

The owner's first instinct was a scheduling app for everyone. Working through it, the pieces separated: engineers' daytime work belongs in the job-management software, where it already lived; the office rota was stable and small enough for a sheet; and the on-call rota was a fairness problem, like the locksmith's. The one real gap was on-call swaps, which happened by text and occasionally left a night uncovered because two engineers each thought the other had it.

The fix was modest: the on-call rota moved into a shared calendar, where each night is an event assigned to one engineer, and swaps are made by reassigning the event, which notifies both people. A monthly AI-assisted fairness check, like the locksmith's tally, keeps the nights even. Total cost: nothing extra. The lesson is to name the specific problem before buying a general tool; "we need scheduling software" often turns out to mean "we need swaps to be visible".

Where switching to an app goes wrong

• Rules never entered. An app enforces only the rules you configure. Deputy and others include compliance settings, but they don't know your contracts. Enter maximum hours, rest gaps and qualifications, then test with a rota that deliberately breaks each rule.
• Availability left blank. Auto-scheduling from empty availability produces nonsense. Get everyone to enter availability before the first auto-generated rota, and set a weekly deadline.
• Skills missing. If the app doesn't know who is a designer, it will schedule two sales staff together. Tag roles and skills before the first auto-schedule.
• Two systems at once. The manager keeps the old spreadsheet "just in case", staff check the wrong one, and the app gets blamed. Pick a launch date and retire the sheet.
• Trusting the draft. Auto-scheduling is a draft too. Review it before publishing, especially the first few weeks, and especially weekends.
• Rules that depend on where you are. Rest breaks, maximum weekly hours and rules for young workers differ between places. Take employment advice if you're unsure what applies, and remember the app's defaults may not match.

A four-week trial that settles it

If you can't decide, run both for four weeks. Keep the spreadsheet (with checks) as the official rota for two weeks while trialling an app's free tier or trial in parallel with two or three staff, then swap for the next two. Track four things: manager hours per week on rota and swaps, rota mistakes that reached staff, swap requests handled without the manager, and what staff say at the end. Most small teams know the answer by week three. And whichever you choose, the hours people actually work still need recording; AI time tracking for small teams covers that next step, and building a rota around demand forecasts shows how busier businesses go further.

Rotas, apps and AI: common questions

Can ChatGPT or Copilot build my rota for me?

They can draft one quickly, but general AI assistants are unreliable at hard rules such as maximum hours, rest gaps and who is qualified for which shift. Treat the draft as a starting point and check it with spreadsheet formulas or a second pass asking the AI to list every rule broken. Dedicated scheduling apps use rule engines designed for this, which is why they're more dependable for larger or rule-heavy rotas.

Will staff actually use a scheduling app?

Most will if it replaces something annoying, such as a group chat of swap requests or a printed rota that changes after it's pinned up. Adoption fails when the manager keeps the spreadsheet going in parallel. Pick one launch date, put availability and swaps only in the app from then on, and give people a fortnight to get used to it.

Do scheduling apps handle working-time rules automatically?

Many include compliance settings for breaks, maximum hours and overtime, and some are configured for particular jurisdictions. None know your contracts or local rules unless you set them up correctly, so enter the rules yourself, test them with an example rota that deliberately breaks them, and check with an employment adviser if you're unsure what applies to your team.

Further reads

• How to Measure Time Saved After Rolling Out AI in a Small Firm — Measure the hours a new rota tool actually saves.
• How to Set Up Human Review for AI Work Without Slowing Down — Keep a quick human check on AI-drafted work.
• AI Scheduling for Small Builders: Subcontractors and Deliveries — Scheduling subcontractors and deliveries around site work.
• How Pubs and Bars Can Use AI for Events, Rotas and Stock — How rotas flex around events and busy nights.
• How to Build a Job Costing Sheet in Excel With AI Help — Another spreadsheet built with AI, for job costs.
• When Not to Use AI in Your Business: 9 Tasks to Keep Human — Why final sign-off on some tasks should stay human.
• Can a Small Shop Use AI Without Having an Online Store? — Where AI helps a bricks-and-mortar shop with no website: till reports, being found on Google, shelf talkers, customer messages and a first-week plan.
• Outgrowing Spreadsheets: When to Replace Manual Excel With AI — Seven signs a spreadsheet has become a system, four routes from AI inside Excel to new software, and a removals firm's job board before and after.
• AI Payroll Checks: Catch Timesheet Errors Before Payday — A pre-payroll check that runs exact rules on timesheets first, then uses AI to read notes, spot odd patterns and draft one query per approver before cut-off.
• How to Run Staff Surveys and Exit Interviews With AI Analysis — A 12-question survey, an exit interview script, anonymity settings, a theme-coding prompt with checks, and an online clothing shop's first round of results.
• How to Automate Holiday Requests and Absence Tracking — Build a leave request process that checks balances, protects shift cover and records changes without repeated messages.
• Capacity Planning With AI: Know When Your Team Is Full — Find the role, day or machine that limits your next booking, with a practical weekly capacity calculation and checked AI scenarios.
• AI Route Planning for Small Delivery Fleets and Mobile Teams — How route optimisers split stops between vans and order them, the stop data that makes or breaks the plan, real prices, and three small-fleet examples.
• AI for HR in Small Businesses: 12 Tasks You Can Hand Over Safely — Twelve HR jobs AI can draft or sort for a small team, each with an example and a watch-out, plus the people decisions that stay human.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: pricing pages for Deputy, When I Work, Homebase and Connecteam; Microsoft 365 Copilot Business pricing.