A model should not simply produce a pass or fail result. It should reveal what is compliant, what is missing, what is risky and what needs to change before those issues reach site, argues Dr Mohammad Mayouf ahead of his appearance at Digital Construction Week (DCW).
Compliance checks are great, aren’t they? All those rules, standards, checklists and reports. But here is the uncomfortable question: if compliance is so important, why do we still leave so much of it until the point where change is expensive, disruptive and difficult?
In construction, compliance is often treated as a gateway rather than design intelligence. A model is produced, information is exchanged, decisions are made and only then do we ask whether the design meets the rules. By that point, a compliance issue is no longer just technical. It has become a programme risk, a cost risk and, sometimes, a reputational risk.
Digital tools and AI are presented as the solution. But not all automation is equal. Many workflows still operate as black boxes: data goes in, an answer comes out, and everyone is expected to trust the result. The model passes. The model fails. A warning appears. But why? Which object triggered the issue? Which rule was applied? Which property was missing? Which relationship was misunderstood?
Explain the logic
If we cannot explain the logic, we have not really automated compliance. We have only hidden it behind a smarter-looking interface. The next step for digital construction is not simply more AI. It is more transparent AI.
Emerging guidance such as PAS 1958 reinforces this by helping the sector understand how data and information standards connect across the built environment. That matters because automated compliance cannot sit outside the standards landscape. It needs reliable information foundations, clear data structures and traceable decision pathways.
This means moving from black-box workflows to white-box workflows, where the logic behind compliance decisions is visible, explainable and challengeable. A door is not just a door. It may be part of an escape strategy, a fire compartment, an accessibility route, an asset register and a handover requirement. Compliance depends on context, not just classification.
This is what it means for a BIM model to say “no”. Not to block creativity, and not to replace professional judgement, but to bring compliance intelligence into the design process earlier. A model should not simply produce a pass or fail result. It should reveal what is compliant, what is missing, what is risky and what needs to change before those issues reach site.
Because the future of automated compliance should not be about trusting the machine, but more so about understanding the decision.
Dr Mohammad Mayouf will dive deeper into this topic on day two of DCW on the Information Management Stage.
