What this solves
The pattern is the same across every industry with documentation requirements. A regulation sets requirements. The documentation sits scattered, unstructured, of varying quality. An inspector will one day arrive to check whether it holds up. And nobody internally has a full overview before that happens.
The obvious move is to ask the practice system. But a practice system tells you what people have said they did. An inspection tells you what the documents actually say. A control point isn't satisfied because someone ticked a field, but because the document exists, is of the right type, covers the right period and was filed within the deadline.
The quality management engine reads the documents and answers one question per control point: does the documentation the requirement asks for exist, is it valid on the inspection date, and does its content hold up against the legal requirement?
The answer is one of five: satisfied, deviation, deficiency, requires manual review, or not applicable. Out comes a deviation report with legal basis, quotation and page number for every finding, and a portfolio overview across every unit being inspected.
How we work
The system consists of two parts held strictly apart, and that's the most important design decision in the product.
The engine takes in a set of documents, extracts text and dates, identifies document types by content rather than filename, binds documents to control points and runs the rules: exists, correct period, age, deadline, consistency, scope. None of those rules are industry-specific.
The control points are the other part. They live as text files, one object per point, with the legal basis and the inspector's instruction reproduced verbatim from the requirement list. They are data, not code.
That means a new inspection year, a new regulation or an area being reactivated requires no code change. You write new control points, not a new engine. It's also why the engine solves a class of problems rather than one problem.
Intake is source-agnostic from the ground up. A ZIP, a folder, individual files or a direct lookup against the practice system all go in the same way and are treated alike. We're happy to start with manual upload, because it lets us show that the inspection logic holds before anyone builds an integration.
Security and governance
Three safeguards sit in the code, not in a prompt.
A machine assessment can never upgrade a finding. If the rule engine has concluded a document is missing, no AI reading can turn that into satisfied. The model can only confirm or downgrade. That's the technical guarantee against the worst outcome, which is a system reporting that everything is fine when it isn't.
No assertion without evidence. For a point to be concluded, the conclusion must point to a quotation found verbatim in the document text, on the page given. If the quotation isn't there, the assessment falls to requires manual review. You can challenge every single line in the report by looking it up, not by trusting us.
Nothing is measured against today's date. Every run is given an inspection date, and all age and deadline rules are calculated against it. That makes runs reproducible, and lets you simulate the inspection before it happens.
Where we connect to a practice system, it's read-only. The system never writes anything back, and a test in the codebase enforces that the rule engine itself may have no network dependencies at all.
And one status does more for trust than any of the guarantees: requires manual review is a full conclusion, not an error message. If the classification is uncertain, the format unreadable or the evidence absent, the system says so. Uncertainty never becomes an assertion.
There's a cost to that. A system willing to answer "don't know" gives you more points to review yourself. In return, the points it does conclude on are points you can build on.
Who this is for
Authorised accounting firms are where the engine is delivered and running. The control points for the Norwegian inspection form are written, and the system pulls the engagement list and the documents straight from PowerOffice Quality. If you use a different quality system, talk to us before you count on anything.
Other reporting-heavy industries are next. The pattern shows up in audit, in construction with central approval and health and safety requirements, in healthcare with internal control regulations, in food production with HACCP, in ISO-certified quality systems and in public procurement.
We're honest about what that takes: the control points have to be written by someone who knows the requirements, and they have to be tested against real documents. That's work, not a switch we flip. But it's work in text files, not a new development project.
Do you have a regulation you need to document compliance with, and a volume of documents nobody has a full overview of? Send us the inspection form or requirement list you're measured against. We'll tell you within a day whether the engine fits your problem, and what would be missing.