Bake This

Production planning for bakeries

It thinks like your best baker. At the speed of a computer.

Bake This watches your sales as they happen and tells whoever is at the oven exactly what to bake and when — flavour by flavour, sheet by sheet — on a wall tablet built to be read in one glance from across the kitchen.

You stop guessing the day at 6 AM. The board keeps the case full through the rush and winds production down to an almost-empty case at close — and every number on it can be explained in patties and minutes.

Request a walkthrough See a day with it

Running today as a live pilot in a working patty shop in Atlanta — on an old tablet on the kitchen wall, refined against every trading day.

The board on the wall. One bake, one start time, nothing to interpret.

What it optimises for

Three rules, in the order your shop already runs on

Every bakery balances the same three things. Bake This applies them in a fixed order, so when they pull against each other — and they do, every afternoon — the trade-off is resolved the way you would resolve it.

Never run out

A customer who walks out empty-handed is the most expensive outcome there is. The planner treats a stockout as worse than any leftover, for every flavour, at every hour — including the slow movers.

Almost nothing left at close

Production tapers toward closing time so the case runs down with the day. You set the leftover you are willing to accept — in Atlanta the target is 20 pieces or fewer across the whole case.

As fresh as possible

Every batch in the warmer carries its age. You set a freshness target in minutes, and the planner prefers many right-sized bakes over one big one that sits.

Rule one outranks rule two, and rule two outranks rule three — always. That order came from a shop owner, not from a spreadsheet.

A day with it

What actually changes, from open to close

Before opening

The doors open on a full case

The pre-open bake is yours to define, flavour by flavour, down to a half sheet — and the board shows it before the day's first customer. No more "what do we usually do on Tuesdays?" The person opening doesn't have to remember; the wall does.

The lunch rush

Every bake is sized for the demand it will serve

A sheet that goes in at 12:05 sells from 12:25. Bake This plans for the moment the bread comes out of the oven, not the moment it goes in — using your shop's own hour-by-hour, flavour-by-flavour history. The board also shows how today is running against the day the plan was built from, so a hot day is visible while it is happening, not in tomorrow's report.

Mid-afternoon, running behind

One glance says what to bake now

When the kitchen gets loud, the board stays simple: the next bake that matters, its sheets per flavour, and when to start it. If a bake should have started already, the board says so. And if a flavour does run out, every request you couldn't serve is counted and stays on the screen — a miss is measured, not shrugged off.

And when you disagree with the plan, you win: override it in a tap and the board follows you.

The last hour

The wind-down to an almost-empty case

From a set time before closing, every advised bake shrinks along a straight line toward zero. The board will never tell anyone to start a bake that comes out of the oven after the doors are locked. The case runs down with the clock, and what is left at close is what you chose to accept — not what happened to survive.

How it decides

No black box. Every number has a reason you can say out loud.

We could call this AI. We'd rather tell you exactly what it does — because when a baker asks "why five sheets?", "the algorithm said so" is not an answer you can run a shop on.

Your history is the forecast

The starting point for any day is the same weekday last week, hour by hour, flavour by flavour — your shop's own sales, straight from the till, not an industry average.

Your judgment is the dial

You set one factor for today — rain coming, a match, a festival, a hunch. The app shows the weather and suggests; you decide. Everyone on shift can be told why today is planned at 80%.

Real ovens, real minutes

The plan respects your equipment: sheets per oven, pieces per sheet, bake time, your opening hours per weekday. It never advises a load your kitchen can't physically produce in time.

Every number has a why

The plan is built only from things you can point at: which day it came from, what is in the warmer, what is already in the oven, what buffer was added. So when a baker asks "why five sheets?", the answer is patties and minutes — never "trust the model".

Every dial is yours

Freshness target, buffers, the closing taper, the pre-open load, per-flavour caps, opening hours weekday by weekday — dozens of dials an owner can change tonight, from a phone, without a developer.

The board needs no login

The baker's screen is a kiosk: mount the tablet, open the board, done. Settings and money numbers stay behind owner and manager accounts.

The answer to "why five sheets?" — as the owner sees it.
Today's judgment, set from the couch the night before.

One shared picture

Everyone sees the same day

The whole chain is on screen — what is in the case, what the till is selling, what comes out of the oven next — so the kitchen and the counter stop working from two different guesses.

The case

Warmer stock per flavour, with the age of every batch. Damage and corrections are recorded in two taps, right where they happen — so the count on the screen is the count behind the glass.

The till

Every Toast sale is deducted the moment it rings up, so the board's stock moves with the queue at the register — not with last night's spreadsheet.

The handover

The counter sees what the kitchen has promised: when the next batch lands, and what went unserved today. "When is the next one out?" gets an answer from the wall, not a shrug toward the oven.

The counter crew gets its own read-only board, and it talks in units, not dollars — patties in the case, patties coming, patties missed — so the whole team can read the day while money numbers stay behind the owner's login.

And the board is honest about itself. If a bake should already be in the oven, it says so. If the till connection drops — or the till is up but no sales are coming through — it warns you, in plain words, to count before trusting the stock, rather than showing a confident number it cannot back.

The Atlanta pilot

One shop first, on purpose

Bake This is not a demo. It runs today as a live pilot in one patty shop in Atlanta — and that shop has shaped every screen on this page.

The algorithm was not designed in a meeting room. It started with more mathematics than it has today, and a working kitchen sanded it down: whatever could not explain itself at the oven, in patties and minutes, went. What is left is the owner's own method made precise — the same weekday last week, adjusted by a factor he sets — arrived at from the shop's data and the owner's experience together. And it is still being refined, deliberately, against every trading day.

The load is real. Days at the pilot shop run from around 1,100 to nearly 1,800 patties — 1,508, 1,786, 1,142 and 1,157 on four consecutive July days, straight from the till. That is the scale this planner works at: a full case, a full oven, all day.

And here is the honest shape of the value. A busy Saturday mostly forgives you — demand swallows your mistakes, and everything sells. It is the quieter days, the slow Monday, the rainy afternoon, where habit bakes for the day you hoped for and the bin gets the difference. Limiting waste on the days that don't forgive is exactly where this tool is pointed.

Are the results encouraging? Yes — and we are careful with that word. The pilot measures what the owner actually cares about: what is left at close against his target of twenty pieces or fewer, whether any flavour ran out, and how old each patty was when it sold. Those numbers will be published here once they have been measured properly, not before. What we can say today: the shop keeps choosing to run its day on this — on an old tablet on the kitchen wall, tuned to its own opening hours and its own weekday rush — and every week it teaches us something we fix.

In the owner's words

We are sitting down with the pilot shop's owner for a proper interview — what changed at the oven, what he overruled, what he would tell another owner. His answers go here, in his words, once he has read and approved every line.

Interview slot

Before and after — how the day's bake was decided before the board, and what close looks like now

Owner's own words · approved by him · no paraphrase

Interview slot

The quiet days — what changed in the bin on slow Mondays and rainy afternoons

Owner's own words · approved by him · no paraphrase

Interview slot

Trust — the first week, the times he overruled it, and what he'd tell another owner

Owner's own words · approved by him · no paraphrase

What it needs

What it runs on

  • Your till. Today Bake This reads live sales from Toast POS as they happen — and if the feed ever goes quiet, the board says so instead of showing a stock count it cannot back. On a different system? Tell us in the form — it shapes what we build next.
  • A tablet on the wall. Any tablet with a browser — including the old one in the drawer. The board is a plain web page: no app store, no install, no new hardware to buy. The Atlanta pilot runs on an old tablet, mounted where the baker already looks.
  • A phone for the owner. Settings, the day factor and the evening's numbers work from your pocket.
  • Multiple locations, one account. Built multi-tenant from day one — each shop gets its own ovens, hours, flavours and settings.

Talk to us

See it on your own numbers

A walkthrough is thirty minutes: your flavours, your hours, your ovens, and what the board would have said last Tuesday. No slide deck.

Franchise operator? Mention how many locations — the multi-location setup is a conversation of its own.

Please tell us your name.

Please tell us the name of the bakery.

That email address doesn't look right.

We use these details only to reply to you. They go to our own server — no third-party form service, no marketing lists. Privacy statement.