Product · October 2026 · 5 min
The ontology, explained simply
The value is not in one more dashboard. It is in the model that connects your data. Here is what it does, in practice.
The problem: one thing, five names, five systems
In any organization that buys, the same information lives in several places. The commitment sits in the financial management system or the ERP, the order in the tool used to exchange with suppliers, the contract in a spreadsheet or with a group purchasing organization (GPO), the stock in warehouse software, the invoice in accounting, sometimes months after delivery.
Every system names things its own way. The same exam glove is called “nitrile gloves M bx 100” on an invoice, “GLOVE NITR. SIZE M” on an order, and “Nitrile exam glove powder-free M ×100” in a catalog. The same industrial bearing carries three part numbers depending on whether you look at the plant, the workshop, or the supplier. No one has a single table that says: this product, at this price, from this supplier, for this site, under this contract, delivered on that day.
What the ontology does
An ontology is a data model that connects the objects of a domain and their relationships. At Miles, in Argos as in Talaria, it is the single table from which everything is calculated. Each invoice line is linked to:
- an identified product, matched to common reference data using industry codes (GTIN everywhere; CIP, UDI, or EMDN in healthcare; manufacturer part numbers in industry);
- a supplier and a contract, with its negotiated price and discount thresholds;
- a site or a consuming department;
- a delivery and a stock position.
Once these links are in place, simple questions finally become simple: how much did we pay for this item this quarter, at which sites, and at what price compared with the contract?
Connect before analyzing. Once the data is connected, the analysis becomes almost obvious.
How it is built
At the start, everything arrives as file uploads: native exports, CSV files, invoices. We build a first ontology on a sample of six to twelve months, then the full ontology on twenty-four months of history at go-live. After that, new invoices enrich it every month; every quarter, new sources can be connected, through an API when the source system allows it.
The older the ontology, the more precise it is: every matched label stays matched for good, and the common reference data grows with each client.
One ontology for healthcare, defense and industry
Product, supplier, contract, price, commitment, stock, delivery: these objects exist in a hospital just as they do at a defense manufacturer. A new industry adds reference data, public sources, and analysis rules. Never a different product.
Who owns it
The ontology belongs to you and remains exportable at any time, free of charge, if you leave. Miles keeps only the item mappings and detection rules, which are its know-how. Argos reads your data and writes nothing to your systems.