Digital twins
A digital copy of a process where a decision can be tested before it costs money. What happens if you add a courier, raise the price by 10% or close a warehouse—we work it out on the model rather than on the live business.
Scope of work
point by point.
- A description of the process and its decision points
- Historical data collected and its quality checked
- Simulation model built on the real distributions
- Calibration: the model reproduces last year
- “What if” scenarios with ranges instead of a single figure
- Optimisation: parameters fitted to the goal
- Web interface: sliders, side-by-side scenarios
- Export of the results and the charts
- Model updates as new data arrives
- Training your team to work with the model
Situations
where this pays for itself.
Logistics and delivery
How many couriers are needed on a Friday evening, how to redraw the zones, what one more warehouse would give you.
−18% distance per route
Production load
Where the queue builds up, what a second shift would give, how batch size affects lead time.
Bottleneck found in 2 days
Pricing
How a price change works through to demand, margin and stock, with seasonality taken into account.
A range instead of a single figure
Support headcount
How many operators are needed hour by hour to hold response time, and what an AI agent would change.
A shift roster for the target SLA
Four steps
from brief to handover.
- 01
Framing the question
A model answers a specific question. “Have a look at the process as a whole” is not a question, and we will help narrow it down.
- 02
Data
We gather the history and assess its quality honestly. If there is too little data, we say so beforehand, not afterwards.
- 03
Calibration
The model has to reproduce a past period within an acceptable error. If it does not, we do not compute the future.
- 04
Scenarios and interface
You get an interface where your own managers turn the parameters and compare the options.
What we
build it with.
Three tiers.
The exact estimate follows the brief.
from ₽350,000
15–30 working days
- One process and one question
- Model built on historical data
- Three “what if” scenarios
- Report with the conclusions and charts
from ₽900,000
30–60 working days
- Several connected processes
- Web interface with sliders
- Optimisation against a stated goal
- Model updates from fresh data
- Team training
on request
60–120 working days
- A model of the whole chain
- Real-time connection to your systems
- Forecast compared against actuals
- Deployment inside your own network
- A year of support and recalibration
Prices are the lower bound. What pushes an estimate up is set out on the pricing page
About this
service.
01How is this different from a forecast in Excel?
Excel works with averages: average demand, average delivery time. Reality lives in the deviations—queues form at the peak, not on average. A simulation model plays through thousands of versions of the day and shows not one figure but a distribution: in 80% of cases you will cope, in 5% you will fail in this particular way.
02How much data is needed?
As a rule, a year of history on the key events: orders with timestamps, operation durations, failures. With less history a model can still be built, but on expert estimates—and we state plainly in the report which conclusions rest on them.
03How far can the model be trusted?
Exactly as far as it reproduced the past. We always show the calibration error: the model is run over last year and compared with what actually happened. If the gap is more than 10–15%, we say so rather than tuning the result to fit.
04Who will use it after handover?
Your managers. That is why the Business package includes not a script but a web interface with sliders and scenario comparison, plus two hours of training. A model only the contractor can use is a useless model.
Get an estimate:
Digital twins
The brief takes 5–7 minutes. In working hours we reply within two hours, and the estimate is free.