Security and trust
We handle data that reveals a network's turnover, its suppliers and its purchase prices. Below is what happens to that data — and what never does.
How data access works
Each network has its own tenant: its own database, keys and schedules. One client's request cannot technically reach another client's data.
A connection to the accounting system starts read-only. Write-back is enabled by a separate client decision, not at deployment time.
Who edited an order, who sent it, to whom and when. The log cannot be edited from the interface — otherwise it would be worthless.
A public read key is issued and revoked by the network owner. No key grants write access.
Responsible AI
The rule is simple: a model does not answer with money, so it does not get to decide with money.
A recommendation always shows the quantities behind it. Only what the network owner has explicitly permitted may proceed silently.
When data is insufficient the system says “not enough data”. That is inconvenient and correct: a convenient number here would be an invention.
An edit is stored with its author and time, and it shows where the calculation systematically diverges from reality.
What we never do
- We never pass one network's data to another — anonymised extracts included.
- We never sell data or use it for third-party market analytics.
- A supplier sees nothing beyond the order addressed to them.
Show us one critical process. We will show how it runs here.
We look at your cycle: how an order is assembled today, who decides, where time leaks and what the system takes over.