Chains of 5 to 20 shops · One tax ID
Stop managing the bakery. Start predicting it.
The central obrador and the shops stop coordinating over WhatsApp. The engine learns every product in every shop, then says what to bake on Monday, what to send to each neighbourhood and what to order on Thursday. On your current POS, from day one.
On your current POS · Data hosted in the EU · Fiscal compliance built in (Spain pack live)
The launch with no history
The roscón you haven't sold yet, forecast from the bread you have.
A seasonal product's first year has no history: a roscón, a panettone, a neighbourhood edition. The engine launches it from an analog product —same family, same seasonality, same price band—, takes its curve and shows you which one it learned from, with a similarity score. As the real sales arrive, the forecast leans less on the analog and more on the product itself.
- Launch forecasting by analog
For a new or seasonal product with no history, the engine picks the closest established product. It takes that product's demand curve with its calendar and its weather sensitivity, scales it, and tells you which one it learned from. The launch stops being baked by guesswork.
- In-shop bake-off window, hour by hour
In shops that finish frozen or par-baked product, the engine lays out the day: when to pull from the freezer, when to proof, and which oven slot each item takes. The oven's capacity and the demand forecast for each hour bound the plan.
- The obrador → shop loop, told in one line
Yesterday's waste at each shop becomes today's suggestion: “18% of croissants left over in Chamberí yesterday → 40 fewer today”, with the reason already worked out. It closes the circle between what was left over and what gets baked.
Daily life in a growing chain
Why does the seventh shop feel three times harder than the third?
Because seven shops are not one shop seven times. Each one has its own street, its own weather and its own hours, and the decisions multiply while the information stays in one head. Four problems appear, and all four come from the same place: nobody knows what tomorrow will bring.
- The Master Spreadsheet
Forty-seven tabs and three colours of conditional formatting. It lives on one laptop. One person updates it, and that person has not taken a real holiday in eighteen months. The day shop eight opens, it stops reconciling.
- The Sunday-night WhatsApp
Nine on a Sunday evening. You message the managers, thirty replies come back, somebody never plays the voice note, and by midnight you have a number you already know is wrong. Every Sunday: 12 to 24 hours a month of the operations lead, purely on coordination.
- The P&L that arrives 15 days late
Your accountant closes the month and hands it over three weeks later. By then the shop that was losing money has been losing it since before the report existed. The quarter's decisions were taken in the dark.
- The afternoon waste
Seven in the evening. The counter in the good shop is bare and the one down the road has whole trays that will not sell tomorrow. Nobody decided this: it was baked just in case, because nobody knew how many people would walk in. That bread is paid for twice — the loaf left over and the loaf that ran out.
All four are the same problem seen from four places: decisions made from memory, late, on last week's data.
Four villains, each with a name.
The inversion
Bake just in time, not just in case.
That is what the BakeWise Biga Engine does. It learns every product in every shop — its weather, its holidays, its neighbourhood. That forecast becomes the bake plan, the delivery route and the supplier order. The plan lands before the weekend, a named person approves it with a signature, and nobody types it again.
In a chain with clean sales history, that usually shows up as less afternoon waste and fewer gaps in the counter at peak hours. Exactly how much is what the diagnostic measures on your own data, before you sign anything.
This is the product. Not an illustration.
The screens an operations lead uses every morning.
Today's Plan
Tuesday 14 October
12
Batches to bake
1.340
Units to produce
4
Deliveries to dispatch
2
Purchases to approve
What do I bake today?
- Rustic baguette620 unitsIn progress
- Butter croissant480 unitsScheduled
- Sourdough loaf240 unitsScheduled
What do I ship today?
- Pozuelo Estación · 410
- La Moraleja · 360
- Majadahonda · 300
- Somosaguas · 270
Departs 06:40 · last drop ~09:15
What do I buy today?
- Harinas del NorteT65 flour · 12 sacks€1,240.00Suggested
- Lácteos Ribera82% butter · 40 kg€486.00Suggested
What to bake, ship and buy today. On one screen, shop by shop.
A reproduction of the real BakeWise interface, with example data.
Six jobs that stop being yours
Which jobs stop being yours?
Six that do not scale with shop count: the forecast, the shared process, stock and traceability, delivery, shifts, and the month-end close. The engine does each one and hands you the result to approve. Nothing executes until a named person signs it, and every plan carries the reason behind it.
Coordinate the network. Don't type it.
| Job | What the engine does | Plan |
|---|---|---|
| What to bake tomorrow, shop by shop | One model per product per shop: your own history, the national weather service, public holidays and neighbourhood patterns (Spain: AEMET, live). By Friday you already know Monday. | Included in Chain |
| The same process in all seven shops | Draw the flow once, publish the version, and every shop runs that one. Head office locks the stages that define the brand and opens the ones each shop may adapt. | From Chain |
| Stock, expiry and traceability without a notebook | Real FIFO with state-aware expiry (dough, par-baked, finished). Forward and backward traceability in minutes. Several warehouses. | From Chain |
| Delivery and cold chain that document themselves | Routes optimised with Google OR-Tools on capacity and delivery windows, signed proof of delivery and a temperature record per leg (Spain: ATP transport evidence, live). | From Chain |
| Shifts and clock-ins that survive an inspection | Geolocated clock-in per shop, inspection export, and an offline mode that does not lose the clock-in. | Included in Chain |
| The month closed without waiting for the accountant | Gap-free numbering guaranteed in the database, delivery notes consolidated at month end, and per-shop P&L. Veri*factu or SII depending on your regime; TicketBAI built and XAdES-signing, with foral homologation in progress. | From Chain (per-shop P&L) |
The sum against a regional manager
What does it replace, in euros?
Not another piece of software: a person. A loaded regional manager costs €42–55K a year, and past five shops you need more than one for real coverage. That is the honest comparison: the salary you would otherwise pay, set against €399 a month and the work the engine already does.
What you already pay, with and without an engine.
| Today | With Chain (€399/mo) |
|---|---|
| Sunday night pinging managers on WhatsApp to confirm Monday's production. | Production plan generated and approved on Friday. Your Sunday is yours again. |
| Per-shop P&L arrives 15-21 days late. Decisions in the dark. | Per-shop P&L up to date. The shop that is losing money, you know today, not in three weeks. |
| Waste estimated by eye, with every shop doing its own sums. | Waste measured per product and per shop, comparable across sites and against the forecast. |
| Seven orders to three suppliers, no negotiating power and the admin done twice. | Aggregated demand, consolidated orders and real volume discounts. |
| Regulatory compliance with spreadsheets and hope. | Auditor-ready exports in two clicks. Gap-free numbering guaranteed by the database. |
| Every shop orders when it can and the bakery finds out at 23:30. | The order cutoff is enforced by the server; the exception is signed by whoever authorises it and lands in the audit record. |
| The shop visit, in a notebook and in the memory of whoever went. | Every visit with its score, its photos and its corrective actions, comparable across shops and over time. |
The sum is done against the cost of scaling the team, never against the price of another piece of software.
It calculates. It warns. It proposes. It does not decide.
No suggestion executes on its own: the engine proposes each shop's volume and a named person approves it with a signature before a production order exists. Everything is traced and everything can be undone.
The first step
How do we start without signing anything?
You give us the sales history, read-only. We measure the real quality of your data, train without seeing the last 90 days and predict that window blind. Then we put avoidable waste, coordination hours and purchasing drift into euros, with the assumptions in plain sight. Four weeks, no cost.
Four weeks. No cost. And if the number does not add up, we stop.
- Your data readiness score
Out of 100, with the published rubric and the concrete list of what is missing and what is wrong. It is yours whether or not we continue.
- The blind backtest
What the engine would have predicted across your last 90 days, product by product and shop by shop, with the methodology written down so your team can audit it.
- The business case in euros
Calculated on your own history, not on a sector average. A document operations and finance can sign — or reject with data.
Read-only access. We install nothing on your network. You change no system.
What it takes to start
Three requirements. Not one more.
If you do not meet them, it is not the moment yet — and we say so before spending a meeting.
- A POS with three months of history
Square, Toast, Lightspeed, Revo, Storyous, Epos Now — or any POS that exports to CSV. We do the import.
- One person on it for four weeks
About five hours a week during setup. After that, one or two hours a week of oversight.
- A line to your supplier or accountant
To reconcile delivery notes, invoices and fiscal regime. Our team runs the setup with you.
What does NOT change
What does not change?
Almost everything. Your POS keeps taking payments, your team keeps working the way it works, and your accountant keeps receiving the same documents. There is no migration and no cut-over day: the engine connects on top and, for the first weeks, only watches and learns. Your recipes stay yours.
Nothing is replaced. No migration. No cut-over day.
- Your POS stays your POS.
- The engine connects on top. You change no system and you do not stop taking payments for a single day.
- The team keeps working the same way.
- For the first weeks the engine observes and learns. Nobody has to touch anything new for it to start being useful.
- Your accountant keeps receiving the same thing.
- The engine generates the delivery note and the invoice, with the numbering and per-rate VAT; your accountant receives them as always.
- The recipes are yours.
- We do not publish them, we do not sell them and we do not train anyone else's models on them. EU hosting.
- If you pause, you lose nothing.
- Full export as CSV/JSON at any time. No lock-in.
- Guided setup is included (4 weeks).
- Our team imports the history, configures recipes and process templates and trains the team, onsite or remote. Day 1 you are already operating.
Built for the chain with a central bakery, five to twenty shops and a single tax ID. What this page says opens in the live demo, not in a list of names.
Price
Against the salary you will not pay.
€399 a month for a chain of 5 to 20 shops. The comparison is the €42–55K/yr regional manager, not the price of another piece of software.
Chain for 5-20 shops under one tax ID. Group when each shop has its own.
See full pricing and comparison →Free audit
Spain Ley 1/2025 — Audit your compliance in 90 seconds.
The prevention plan has been compulsory since 2 April 2026. Ninety seconds to see what you already have and what is missing.
Questions a chain asks
How long does the rollout across 7 shops take?
Four weeks of guided setup. We start with two pilot shops and your team does not change its routine while the engine learns. The other five are brought in parallel with already-calibrated data. Without stopping operations in any of them.
Does it work with my current POS?
If your POS exports sales to CSV, has an API or drops files on an SFTP, yes. Native connectors ship for Square, Toast, Lightspeed, Revo, Storyous and Epos Now. For anything else (hioPOS, Vantas, TPV Fácil and the rest), setup configures the CSV adapter against the format we publish. It works, but it is a delivery route, not a plug-and-play native connector.
Does it read my POS and my ERP?
We work with your ERP, we do not replace it. Tills are read through their own connector or a CSV/SFTP route; ERP systems through their API or their published export. Accounting, invoicing and payroll stay where they are.
Do we have to train the whole team?
No. We train one admin per shop (1-2 hours) and the operations lead (3-4 hours). The rest of the team works from their phone with no specific training: the logic is WhatsApp or Google Maps.
What happens to my recipes and my data?
They are yours. EU hosting. We do not use your recipes to train global models: models are calibrated with your data and used only for you. Full export (CSV/JSON) at any time. No lock-in.
I run a central obrador + 6-20 shops under one tax ID. Do I need Group?
No — unless you want separate fiscal identities. Chain (€399/mo) supports up to 20 shops under one tax ID, central obrador not counted. That includes optimised routes, the driver app, cold chain, per-shop P&L and the whole operations stack. You only need Group when each shop has its own tax ID (franchises, multi-brand, several companies): there each child tenant invoices separately under its own regime.
Franchise or own-store chain? Chain or Group?
It comes down to the tax ID. If it is one company with N owned shops (one tax ID, unified invoicing), Chain (€399/mo). If it is a franchise network with a tax ID per franchisee, Group (quoted). Each shop is a child tenant with parent-child hierarchy, separate invoicing and its own fiscal regime per entity. Half an hour of conversation and we tell you which one fits.
What does Chain guided setup include?
Four weeks with our team, onsite or remote: POS history import, recipe and process-template configuration, team training (one onsite session plus materials) and a 30-day review with the return report. Day 1 you are already operating and your team touches nothing during the import.
Are the AI production suggestions applied automatically, or do they need approval?
Approval is always required. BakeWise proposes — it never decides. The ProductionAgent generates daily volume suggestions per shop with status 'pending approval'. The shop manager or the responsible lead reviews and approves with a signature before production orders are created. No suggestion materialises automatically. You can configure which suggestions need explicit approval and which are accepted by default — always with complete traceability.
Can a shop manager see everyone’s P&L?
Only if you let them. From Chain, each person can carry a department — production, procurement, quality, finance, people — and their permissions resolve from that role + department pair. It switches on person by person: assigning a department is what restricts, and anyone without one keeps working exactly as before. Nobody gets locked out on the day you turn it on.
With Spain's Ley 1/2025, do you handle surplus-donation agreements?
Yes. The waste-prevention plan comes out of your own waste and production cycle. On top of it you keep the register of agreements with each receiving body —food bank, NGO, surplus marketplace— with its validity. The engine warns you before an agreement expires, so you have time to renew it. The warning lands in the decision queue, not in an email nobody reads.
Your next shop should not be three times harder than the last.
Start wherever you like: open the demonstration chain right now, or request the diagnostic and decide with your own numbers.
Shop-by-shop rollout, without stopping operations. You start with two and scale the rest.