Contextualization: your data described in your company's terms.

We define with you what a client, a project or a margin is in your company. Your reports, your agents and your AI assistant use those definitions.

What it is for

Once centralized, data is still raw: inconsistent formats, duplicates, missing fields. The word "client" does not always mean the same thing in the CRM and in accounting.

A general-purpose AI model does not know your business rules. We write these definitions with you, and every report and every agent then relies on the same ones.

This method comes from our data work with SMEs since 2021: how we got here.

Illustrative workflow

One definition of margin

Finance and operations may include different costs in their calculation. We agree with you on what to count and which sources to use. Dashboards, agents and the assistant's answers then use that definition.

What we put in place

Beneath that definition, what we build with you so your data means the same thing everywhere.

Data transformation

We convert, normalize and deduplicate raw data so it can be compared from one system to another.

Data model and business context graph

We structure the data around your business entities (clients, products, contracts, orders) and the links between them, with their source and date. The definitions are written as code and versioned, and each term keeps a single meaning.

Integrity and governance

We check data quality, trace where data comes from and document the business rules that apply.

Frequently asked questions

What happens when a definition changes?

We update the rule agreed with your team and check the reports and agents that depend on it.

Who owns the context?

You do. Your models, your definitions and your context graph can be exported in open formats.

Which data can be contextualized?

Data that describes your business entities or events: clients, products, transactions, contracts, communications and internal documents. Technical logs stay in their raw format.