// SERVICE · DIGITAL TWINS
A working model of your production — that plans your week.
// Forecast. Simulate. Plan. In your own currency.
We build a digital twin of your business — demand, production, capacity, raw materials and cash — calibrated on your own history. It forecasts the next 8–13 weeks, prices every what-if before you commit, and every Monday hands you the plan: what to make, what to buy and by when, and what is about to go wrong.
Start with a data diagnostic: send three exports, get a report in five working days.
// production twin · live demo
Meet the Production Twin.
A control room for a business that has never had one. Every tank, roaster and pallet on one screen; three numbers in money at the top; and a what-if you can drag. Click a unit, move a slider — the whole plant answers.
- Cash above the norm 0 k €
- Wholesale on time 0 %
- Write-offs 0 %
- Capacity used 0 %
- Active alarms 0
drag to pan · ctrl + wheel or pinch to zoom · double-click a unit
An abstract plant on synthetic data. Your twin is built on your own records.
// the monday message
The plan comes to you. No dashboard to dig through.
Every Monday at 9:00 the twin sends the week’s production plan, the purchase list with order-by dates and the warnings that matter — to Telegram and the dashboard. Ask a what-if and get the answer in money. (An example from a coffee-roastery model.)
- Cognitify Twin: Good morning! The plan for week 41 is ready: 14 roasts on 2 machines, 1.9 t of green coffee, capacity at 92%.
- Cognitify Twin: ⚠ Brazil Cerrado runs out on Thursday. Order 360 kg by Tuesday to keep wholesale on time.
- Cognitify Twin: ⚠ Friday doesn’t fit: 3 batches over capacity. Options: a Saturday shift, +2 hours on Thursday, or move the café blends to next week.
- Cognitify Twin: ⚠ 42 kg of Ethiopia on the café shelves is past 21 days — worth pushing this week.
- You: What if the Brazil price goes up 25%?
- Cognitify Twin: House-blend margin drops from 38% to 31%. Buying 600 kg now instead of 360 kg locks in today’s price but ties up €4,100 more for five weeks. Both plans are in the dashboard.
- Scenario saved · Brazil +25%
Your spreadsheet records what happened. A twin tells you what happens next — in money.
Most producers plan the week in a spreadsheet and in one experienced person’s head. It works — until demand spikes, a supplier slips three weeks, or that person goes on holiday. Then cash freezes in raw materials nobody needed, wholesale orders ship late, and product quietly ages on the shelf. Knowledge that lives in one head isn’t a system.
A digital twin is a working model of your business, calibrated on your own records. Machine learning forecasts demand; simulation computes the consequences; a planner picks the plan. Forecasts come as a range, not a promise — and simulation and planning are classical, reproducible maths: the same inputs give the same plan, and you can see why.
// what it moves
Three numbers, in your currency.
- 01
Cash frozen in stock
How much money sits in raw materials and finished goods above what the plan actually needs — and when it can come back.
- 02
Orders on time
The share of wholesale and B2B orders that ship on the promised day — and which ones are at risk this week.
- 03
Write-offs
What expires, spoils or ages out before it sells — caught weeks ahead, not at the stocktake.
// what it does
Six things the twin does every week.
- 01
Demand forecast
Demand 8–13 weeks ahead per product and channel — wholesale, your own outlets, the web shop — as a range, not a single guess. Seasonality, holidays, weather and promotions included.
- 02
Production plan
A weekly plan that fits your real capacity — machines, tanks, shifts, changeovers. When the week doesn’t fit, it says so and offers fixes: an extra shift, two more hours, another site.
- 03
Purchasing with order-by dates
Reorder points for every raw material with supplier lead times built in — a purchase list that says what to buy, how much and by which date.
- 04
What-if scenarios
A key ingredient up 25%. Two new wholesale accounts. A second machine. A supplier three weeks late. Currency swings, power outages. Each scenario runs hundreds of times and comes back priced in cash, service and margin.
- 05
Early warnings
What runs out, what won’t fit the week, what is ageing on the shelf, which customer is sitting on your returnable packaging — flagged before it costs you.
- 06
The money view
Cost per unit sold, not just produced. Margin by product and channel. Cash tied up, week by week. The numbers your accountant sees at month-end — every Monday.
// how we start
From three exports to a working twin in weeks.
- 01
Data diagnostic
Send three exports — sales, production, purchases. In five working days you get a report: what your data already shows, where money is stuck, and where the model would be weak.
- 02
A twin on your data
We calibrate the model on your history, map your capacity and suppliers, and check it against weeks you have already lived through.
- 03
The weekly plan
The Monday plan and warnings go live in Telegram and the dashboard. The few numbers only people know, your team enters with a short message.
- 04
Grow
Connect your web shop, POS and accounting, add sites and scenarios — once the first loop pays for itself.
// built to fit
No sensors to buy and no ERP to replace. The twin starts from the spreadsheets and exports you already have — CSV or Excel, in any language — and connects to your web shop, POS and accounting as you grow. Hosted by us in the cloud, or on your own server if your data must not leave your network.
// who it’s for
For producers who batch, stock and ship.
- 01
Coffee roasters
- A weekly roast plan across machines and blends.
- Green-coffee reorders with order-by dates.
- Freshness on café and retail shelves, tracked.
- 02
Craft breweries
- A tank calendar that warns two weeks ahead.
- Cost per litre sold, not just brewed.
- Kegs sitting at customers — counted in money.
- 03
Food & drink producers
- Plan batches against shelf life.
- Cut write-offs on seasonal lines.
- Hold less raw stock without missing orders.
- 04
Small manufacturers
- Match production to demand, not gut feel.
- See when a second machine or shift pays back.
- Stress-test supplier delays before they happen.
// questions
Questions, answered.
- Is it a fit for us?
- Usually, if most of these are true: expensive raw materials with long lead times; batch production on limited capacity; a perishable or seasonal product; many small B2B customers plus your own outlets; and records you can export.
- Do we need sensors or IoT?
- No. Most of what the twin needs is already in your sales, production and purchase records. The few numbers only people know are entered with a short message.
- Do we have to replace our ERP or accounting?
- No. It reads exports from what you run today — accounting software, spreadsheets, your web shop — and direct connectors come later, when they are worth it.
- How accurate is the forecast?
- Accurate enough to plan with, and honest about the rest: you get a range, not a single number, and the diagnostic shows where your data makes the model weaker. The value is in the decisions — cash, on-time orders, write-offs — not in an accuracy score.
- Is it a black box?
- No. Machine learning is used for the forecast only. Simulation and planning are classical, reproducible methods: the same inputs give the same plan, and every number can be traced.
- Where does our data live?
- Hosted by us in the cloud, or on your own server if it must stay inside your network. It is never used to train anyone else’s models.
- What does it cost?
- It starts with a fixed-price data diagnostic, then a monthly subscription sized to your business. We quote after the diagnostic, not before — so the price matches your data, not a guess.