An AI context engine gathers, links and keeps current the information an AI agent needs to act correctly in a particular business: the records, how they relate, the rules that apply and what happened before. For each task it supplies just the relevant context, with links to the sources.
Why agents need it
A language model knows a lot about the world and nothing about your company: which customer has a discount, which supplier needs checking, or that an urgent email refers to an order already in the ERP. Without context, an agent guesses.
Not search, not a warehouse
- Search finds documents. A context engine connects the records about a thing.
- A data warehouse reports on the past. A context engine keeps working context current, including email and documents.
- A knowledge base holds what people wrote down. A context engine links it to what the systems show.
What it holds
Entities (customers, orders, invoices), the relationships between them, how processes really run, the rules that apply, a history of changes, and where each fact came from.
Context can go stale and links can be wrong, so provenance and human review matter. Read what we mean by shared context or how the platform works.