AI is doing something clarifying to the design and engineering profession. It is separating two kinds of work that have been bundled together for decades, and as an owner, understanding which one you are actually buying has become a meaningful competitive advantage.

The two lanes

Call the first one the drive-thru. Code checks, standard details, redlines on typical conditions, takeoffs, specification sections for familiar building types. This is work that follows rules, repeats across projects, and can be templatized. AI handles it well. The inputs are known, the outputs are checkable, and the quality standard is compliance, not craft.

Call the second one the chef’s table. Unusual site constraints. Structural conditions without clean precedent. MEP sequencing on a phased renovation with a live facility operating throughout. Multi-stakeholder trade-offs where the answer depends on judgment built from hard experience. This is work that cannot be templatized because no template exists. AI assists but does not lead. Human judgment is the product.

The honest problem: most firms present everything as chef’s table. The fee structure treats everything as if it requires bespoke expertise. But a significant portion of what goes into a standard commercial project is drive-thru work: repeatable, rule-based, fast to produce once the right tools are in place.

Why this matters to your budget

When drive-thru work is priced like chef’s table work, you are paying a premium for craft that is not being applied. The standard detail did not require thirty years of experience. The code check did not involve genuine ambiguity. The specification section for an office tenant fit-out was not a novel intellectual contribution.

AI makes the cost difference between these two categories of work increasingly visible. The drive-thru tasks that used to take four hours now take forty minutes. If the firm is not passing any of that efficiency to you (through lower fixed fees, faster delivery, or more scope for the same price) you are funding their margin improvement.

The firms being honest about this are starting to price explicitly by work type. Templated scope at templated prices. Novel problems at premium prices. That is a deal worth making, because you know exactly what you are paying for.

Where projects actually fail

The failures do not live inside either lane. They live at the handoffs between them.

The standard structural detail meets the unusual MEP coordination condition. The code-compliant design hits a permitting authority with a non-standard interpretation. The templated schedule assumption collides with a site condition that was never modeled.

That is the seam. And the seam is where judgment matters most, because the problem is not contained within any single discipline’s scope. It lives in the space between the structure and the MEP, between the design intent and the field condition, between what was assumed and what is actually true.

The project teams that manage seams well catch them early and small. The ones that do not catch them late and large.

What to ask before design begins

Two specific questions worth raising at project kickoff.

First: which parts of this project are standard-scope deliverables, and which require genuine problem-solving? Ask your design team to explicitly separate those categories and price them differently. If they push back, if everything is custom and novel, press harder. Almost no project is 100 percent chef’s table.

Second: how are you managing the interface between disciplines, and how early? The most expensive problems in construction originate at discipline boundaries. The teams that review interfaces formally at 10 percent, 30 percent, and 60 percent of design completion catch the seams before they become field conditions. Ask what that looks like on your project and who owns it.

The firms that give you specific, operational answers to both questions are the ones who have thought carefully about where complexity actually lives on a project. The ones who answer in generalities, “we coordinate closely throughout”, are the ones whose coordination problems will end up in your change order log.

Label the lanes. Own the seams. Your project runs better when everyone is clear about which is which.