Cintoo, the Netflix for 3D Data with Dominique Pouliquen

Most project teams now sit on more scan data than they have ever had. Laser scanners got cheaper, mobile mapping arrived, and 360 cameras clip onto every hard hat. Yet the same teams still struggle to answer a plain question quickly: does what we built match what we designed?

That gap, between a clean digital model and messy site reality, is where budgets quietly leak. Closing it is less about capturing more and more about making what you capture usable, shareable, and fast to act on.

The value of reality capture is not just the 3D picture; it is a faster, shared answer to whether what got built matches what was designed.

Sell Access to Reality, Not Seats

At Cintoo, an early pricing decision shaped everything that followed. Rather than charge per user, the platform let customers invite unlimited people, contractors and clients included, and billed instead on the volume of data uploaded.

A first attempt at consumption-based pricing on tokens backfired. In large firms the buyer and the user are rarely the same person, and users hesitated to spend what felt like the company’s cash. So the model shifted to something legible: subscribe to the number of scans you have, upgrade when you need more.

The effect was a much wider user base, because nobody was rationing access to protect a token balance.

When pricing punishes the people you most want using the product, adoption stalls no matter how good the technology is.

A Digital Twin Might Just Be Your Scans

The term digital twin came out of energy and heavy industry, where a single facility can hold six figures of tagged equipment. The instinct was to remodel all of it into a clean CAD or BIM model, then connect that model to sensors and maintenance systems.

In practice, remodeling a facility that has no current drawings is slow and expensive. A more pragmatic view is gaining ground: if the scan data is segmented so each pump, valve, or pipe can be isolated and tagged, the scan itself can be the digital twin. The information already lives in the capture.

The cheapest twin is often the one you already scanned; remodel only what the scan genuinely cannot tell you.

There Is No Scan-to-BIM Button

The dream is a single click that turns a point cloud into a finished, level-400 model. It does not exist, and pretending otherwise sets teams up to fail.

What does work today is narrower and still valuable. Machine learning can now recognise fifty-plus classes of equipment and segment pipes and walls automatically. But linking a detected valve to its exact serial number, or a 3D tag, still needs a person, because the scan is missing context, contractual scope, and intent that no model can infer on its own.

Treat AI as the fastest way to classify and locate reality, not as a replacement for the judgment that gives it meaning.

Small Errors Are the Expensive Ones

The failures that hurt are rarely dramatic. More often a duct or a pipe sits a few inches off design. On its own that is trivial. Passed downstream to the next trade, it becomes rework, delay, and occasionally a demolish-and-rebuild.

The real product for large firms is not prettier documentation. It is a shared source of truth that lets fragmented teams see the divergence between as-built and design intent early, and decide faster. Colour-coded comparison, green for match and red for deviation, turns a specialist’s private knowledge into something the whole project can read.

Speed of decision against real conditions, not model fidelity, is what separates a controlled project from an expensive surprise.

Practical Takeaways

• Price digital tools so more people can use them, not fewer; tie cost to data or value, not seats.
• Before remodeling a facility, ask whether well-segmented scan data can already serve as the twin.
• Use AI to detect and locate equipment, but keep a human in the loop for serial numbers, scope, and intent.
• Compare as-built against design intent continuously, and surface deviations in colours anyone can read.
• Check where your data physically lives; sovereignty and self-hosting are now real buyer questions.

Reality capture is shifting from an archival record you consult after the fact to an operational feed you act on daily. As continuous capture becomes normal, the advantage will belong to teams who can compare today against yesterday and move, not just those who can scan.