Originally written in French. Translated by AI — the meaning has been preserved, not the prose.
Short definition
An explicit, structured description of what a domain is made of: the types of object that exist, what characterises them, what connects them, what is always true about them, and the words used to refer to them.
Full definition
The word belongs first to philosophy, then to an established technical field — dedicated languages, engines that infer conclusions from given rules, associated standards.
In the articles on this blog, it refers to applying that kind of description to a piece of software, in a deliberately bare form: one text card per business object, linked to the others. Applied to a product, it answers five questions — what the product is made of (the objects and their special cases), what characterises them (the properties), how they connect (the links, each named by a verb), what is always true (the rules), what they are called (the words, in every language and in the field). Those five boxes also act as an admission filter: anything that fits none of them has no place in the shared reference.
In practice, people rarely use the learned word: they say "the reference", "the product model", "the cards".
Usage in the field
It comes into play wherever you want to make a product queryable without opening the code: answering support and pre-sales questions, framing a change handed to an agent, deriving test cases, taking stock of translations, triaging a customer report.
Layer added by "Ontology-Driven Design: what if the ontology became the safety net for AI-generated code?" (2026-10-29). The term also refers to the shared reference a coding agent consults before and after a change: extracted from the code by Claude Code into an ontology.md file, the ontology becomes the reference for business meaning against which gaps in the code produced are judged.
Synonyms and variants
Object reference model, domain model, product model, "the cards".
Not to be confused with
- Glossary entry — a note that pins down one word at a time, to make it usable in decisions and deliverables. An ontology contains a lexicon, but the links between objects count as much as the definitions.
- Database schema — the organisation of data according to technical constraints of storage and integrity; its tables do not map one-to-one onto business objects.
- Product documentation — a narrative for a human reader, organised by journey or by feature, with no queryable structure.
- Knowledge graph — a representation of notes or files as nodes and relationships, usable for computation; it connects ideas without describing a domain.
- Gap — a divergence between what the product does and what it should do; it is recorded separately from the description, which describes reality alone.
Examples
A "Sales Order" card: a one-sentence definition, special cases (urgent order, standing order), properties (a status among four values), links (placed by a customer, contains lines), rules ("an invoiced order accepts no further lines"), lexicon (order, sales order, pedido, "deal" among customers).
Ambiguities / debates
For many people the word conjures up the field's formal apparatus, which makes the work look out of reach; used here, it assumes neither a dedicated language nor an inference engine. Conversely, practitioners in the field would argue that a text card with no formalism is not yet an ontology.