From the founders

Button

1/8/26

59% of architecture practices are using AI

AI adoption in architecture is rising fast. Depending on which survey you read, you will get a different number, a different sample, and a different definition of “using AI.”

But the direction is clear.

Sources we’re using:

RIBA reports 59% of practices using AI in 2025.
AIA research on perceptions and opportunities for AI.
NBS Digital Construction Report on AI adoption and digital anxiety.
Bluebeam AEC Technology Outlook 2026 on early adopter ROI and uneven adoption.

The adult questions are not “Are you using AI?”

They are these:

Is it reliable?
Who is accountable?
Does it protect design intent?

If you build for a living, there is a fourth question you ask without even realizing it.

Will this reduce rework, or will it create more work?

Because in AEC, speed is not a personality trait. It is a workflow outcome.

The part of the job AI should touch first

Most AI conversations still orbit the flashy part of the profession. Images. Renderings. Concepting.

Those are fun. They are not where schedules go to die.

Schedules die in the unglamorous stuff:

coordination gaps between disciplines
code misses that show up late
the “final” set that still comes back with comments

If AI is going to earn its keep, it has to show up where rework lives.

Not at the end of the process. Early.

Using AI early

Catches coordination and code issues upstream
Reduces back-and-forth, not just drafting time

That line matters.

Fewer surprises. Cleaner sets. Fewer review cycles.

What the adoption stats are actually saying

When you read the surveys, the story underneath the headline number is consistent.

Teams want the upside. They are cautious about the downside.

That is not resistance. That is professionalism.

Architecture is one of the few professions where mistakes become concrete, and then become expensive.

So when a firm says, “We’re experimenting,” they usually mean, “We want value without chaos.”

A practical playbook for using AI without turning your team into babysitters

1) Pick a bottleneck, not a tool

Start with “Where do we lose time?”

A few common answers:

recurring plan check comments
RFIs tied to coordination
inconsistent information across sheets
missing details that trigger clarifications

2) Put AI where it can scan with high precision

AI is strongest when it is:

checking for completeness
flagging inconsistencies
spotting common patterns
surfacing issues early

AI is weakest when it is:

making judgment calls
inventing assumptions
filling in gaps on real engineering decisions

3) Make accountability obvious

If AI flags an issue, someone owns the decision.

If AI misses an issue, someone still owns the outcome.

The clean approach:

AI supports detection
licensed professionals own decisions
humans review the final set and sign their name to it

4) Measure real results

The right metrics are operational:

fewer plan check cycles
fewer “clarify” comments
less rework per project
more predictable turnaround

If the work gets calmer, you’re winning.

If the work gets louder, you’re paying for noise.

Where Spacial fits

Spacial was built around one belief: AI is most useful when it strengthens engineering judgment, not when it tries to replace it.

So we do two things at once.

1) Use AI to catch issues early across structural, MEP, and energy

Our AI agents help scan plans for common misses, inconsistencies, and coordination problems across structural engineering, MEP engineering, and energy efficiency requirements, before they become redlines and plan check comments.

2) Licensed engineers review and stamp every set

AI supports the process. Licensed structural engineers and MEP engineers own the decisions and the deliverable.

What you should feel as an architect or builder:

one partner for structural engineering, MEP engineering, and energy compliance
fewer handoffs across structural, MEP, and energy efficiency
cleaner submittals, with fewer plan check comments
fewer surprises downstream