What it means
The opposite is a set of products matched by name. The CRM knows the customer, the helpdesk knows someone with a similar email address, and the finance tool knows an account code. In one data graph there is one customer, and the deal, the project delivering it, the tickets raised against it and the invoice for it all point at that one record.
Graph here describes the shape of the data, not a particular database: records connected by real relationships rather than by matching text. Change one, and everything that points at it sees the change.
Why it matters when you choose software
Most useful questions in a business cross two systems. Which projects are over budget for clients with a renewal coming up? Which customers raising tickets are also late paying? On matched records each answer is a reconciliation job. On one graph it is a query.
It is also what AI depends on. A model can only be grounded in records it can reach, and it can only reach across the business if the business is one set of records.
Questions to ask a vendor
- If we rename a customer, does every module see the change at once?
- Are the modules separate databases kept in step by a sync, or one schema?
- Can a report or an automation join sales, support and finance records directly?
- What does the AI read: the live records, or an index built from them?
- When two teams update the same customer, is there one record to update or two to reconcile?
How Cyril approaches it
Cyril is built on one schema. Sales, Projects, Support, Finance, Marketing, Docs, Mail, Drive and Calendar are modules over the same records, under one permission model and one audit log, rather than applications stitched together. Every record also has an ai_context serializer, a view of it built for the model, which is how an answer that crosses modules is grounded in the same data each time it is asked.