Inside buildingSMART - The Future of IFC with Eric Bugeja

Construction has been talking about good information management for more than 20 years. Structured data, consistent classification, governed workflows: the principles are not new. What is new is that AI has removed the industry’s tolerance for getting them wrong.

Every firm now experimenting with AI is about to discover the true state of its data. For some, that will be an advantage. For many, it will be an expensive surprise.

AI does not solve a fragmented data environment. It scales it.

Bad Data In, Bad Decisions Out, Faster

AI systems are only as good as the data they are trained on and given to work with. If your information environment is inconsistent and ungoverned, AI will not clean it up. It will amplify the mess at speed.

In practice, construction has a specific weakness here: the industry rarely reports well when things go wrong. A construction program that failed on site but was never updated looks, to an AI reading historical data, like a program worth repeating. The model will confidently recommend the same mistake.

The counterweight is structure. When data follows open schemas like IFC, an AI knows exactly what a door is and what a window is, and can deliver reliable results in clash triage, program development, and live digital twins.

An AI initiative that skips the data governance conversation is not an AI initiative. It is a way of automating your existing problems.

Vendor Lock-In Is a Business Risk, Not an IT Preference

The case for open standards becomes obvious the moment you ask one question: what happens to your data if the vendor disappears, or quadruples the price?

A useful rule of thumb comes from the procurement stance Eric applied in large firms: any software adopted must provide open access to the data. Not necessarily openBIM-compliant, but never locked in. The 70s and 80s model, where a system held your data hostage for 20 or 30 years, is a risk no asset owner should accept again.

The point of a standard like IFC is durability. A file received today should still open in 30 years, without a specific software version.

Evaluate every new platform on how easily your data leaves it, not just on what it does while you are a customer.

Most IFC Frustration Is a Requirements Failure

Plenty of people are frustrated with IFC. The pattern behind most of that frustration is a client asking for “an IFC file” with no information requirements attached. The designer hits export, includes everything, and delivers a massive file nobody can manage.

What tends to work is the opposite: tell your supply chain exactly which parameters you need, at what level of geometry, exported in what way. One construction company solved its handover problem simply by issuing export instructions to its designers.

Open standards fail most often at the point of specification. Define what you need from the file before you blame the format.

The Blocker Is the Boardroom

People on the ground and in middle management can usually see the benefits of digital workflows. Funding requests then travel upward and die at the executive level, where awareness is thinnest.

That is why executive education may matter more than another tool rollout. The message that lands: this is not a piece of software being implemented, it is a different way of designing, constructing, and handing over, and it fails if even one party in the lifecycle opts out.

The good news is that the change sticks. Engineers who move away from sketching and handing off to drafters almost never go back, including sceptics in their 60s and 70s, once they experience running thousands of design iterations instead of three.

If your digital business case keeps dying at the top, the problem is executive awareness, not the business case.

Practical Takeaways

• Require open data access as a condition of any software purchase, and test how data exports before signing.
• Issue explicit IFC export instructions to designers: named parameters, geometry level, and method.
• Audit whether your project records capture what actually happened, not just what was planned, before feeding them to AI.
• Build a short executive briefing that frames digital ways of working as delivery efficiency, not software.

The industry has spent two decades building the standards framework that AI now rewards. The firms that took structured data seriously are entering this period with a head start, and one of the slowest-adopting industries has a genuine chance to become a leading one.