There’s a children’s book about a construction machine forced to stop working when a caterpillar builds a cocoon on his bucket. In the stillness, the machine notices the world around him. When the butterfly emerges, both have been changed.

The caterpillar doesn’t add wings to its existing form. It doesn’t take a class on flight. It dissolves into something unrecognizable inside the chrysalis and reconstitutes itself as something entirely different.

That’s not optimization. That’s transformation.

And it’s exactly what most firms are refusing to do with AI.

The caterpillar delusion

The pattern playing out across the AEC industry right now is consistent: firms add AI tools to existing workflows, call it innovation, and expect meaningfully different outcomes.

ChatGPT gets licensed. A pilot project gets designated. Someone builds a demo. Leadership posts about being “AI-forward” on LinkedIn. The actual work of producing buildings happens exactly as it did before.

This is strapping wings onto a caterpillar. It looks like something is happening. Nothing is actually changing.

The firms that are genuinely transforming aren’t adding AI to their current process. They’re asking which parts of their current process should exist at all. That’s a far more demanding question. It means some workflows don’t survive. Some roles look different. Some things the firm has been paid for by the hour become things the firm delivers as part of a fixed outcome.

That’s the liquefying part, the part that’s genuinely uncertain. And it’s the part most firms are trying to skip.

Why incremental feels safer (and isn’t)

The argument for incremental AI adoption sounds reasonable: go slow, manage risk, protect your existing business while you test new approaches.

The problem is that the competitive landscape isn’t moving incrementally. The AI-native firms entering AEC aren’t asking “how do we add AI to design-bid-build delivery?” They’re asking “what does building delivery look like if you design the process around AI from the start?”

Those are different questions with different answers. The incremental adopters will end up with better versions of the old thing. The transformers will end up with something that makes the old thing increasingly hard to justify.

And you, as the building owner, will eventually notice the difference: in cost, in speed, in the quality of information you receive throughout the project.

What the chrysalis looks like in practice

For a design firm, genuine transformation means AI is shaping what gets designed, not just how it gets documented. Coordination checking happens continuously, not in milestone reviews. RFI volume drops not because the response process is faster but because the conditions that generate RFIs are caught in design. Deliverables shift from drawing packages to decision-quality data that feeds directly into operations.

For a contractor, transformation means the job site is producing real-time data that flows back into planning decisions, not just generating reports. Self-performed work enabled by robotics changes the subcontractor model. Predictive scheduling replaces reactive problem management.

Neither of these looks like the current business with new software. Both of them require the willingness to let the current business dissolve into something different.

What you should be watching for

You’re choosing design and construction partners for projects that will shape your organization for decades. The firms worth choosing are the ones willing to go through the chrysalis, not the ones promising that you’ll barely notice the change.

Ask your prospective partners what’s different about how they work compared to two years ago. Not what tools they’ve added, what’s different. What has changed about the work itself, the roles, the decisions, the deliverables?

If the answer is “we’ve added some great tools to our existing process,” you’re talking to a caterpillar with borrowed wings.

If the answer is “here’s what we used to do that we no longer do, and here’s what replaced it,” you’re talking to someone who’s been through the chrysalis. They’re messier, perhaps. Less polished in their explanation. But they’re actually different.

The building industry will look different in ten years. The firms helping you build it in ten years will look different too. The ones worth staying close to are the ones choosing to become different now, rather than waiting until they have no choice.