AI for Architects: 5 Ways to Use AI on Real Project Work

Last month, I asked a friend of mine how he was using AI. He’s a principal at a 40-person firm, with 25 years of experience, sharp as a tack, and not afraid of technology. His answer: “I tried ChatGPT a few times. Asked it to rewrite an email. It was okay. Haven’t really gone back.”

I hear some version of this from architects all the time. They’ve kicked the tires. Maybe they’ve asked it to summarize something or draft a paragraph. It was fine. Not life-changing. So they moved on, because they have real work to do.

Here’s the thing: the architects who are getting massive value from AI aren’t using it only to rewrite emails. They’re using it as a thinking partner for the actual work of running projects, managing clients, and producing deliverables. The gap between those two approaches, between those who tried it once versus those who use it daily, is enormous. And it’s only getting bigger.

I’ve spent the last year going deep on this in my own work. Not the flashy stuff like AI renderings or generative design. I’ve been looking at the unglamorous, essential, daily stuff that eats your time. I’ve found some very specific ways architects are using AI right now to get more done with the expertise they already have.

Here are five of them.

1. Pushing back on scope creep without burning the relationship

Every architect has a version of this email sitting in their drafts folder. The client wants something that’s clearly outside the agreed scope, and you need to say no without torpedoing the relationship. You know the tone you want. You can’t find the right words.

This is one of the best “try it right now” use cases for AI. Describe the situation, tell it the tone you’re going for, and you’ll get a draft that’s direct without being adversarial. You’ll probably want to edit it, but you’re editing instead of staring at a blank screen for twenty minutes. That’s the difference.

2. Turning a 200-page report into a one-page brief

A building condition assessment lands on your desk. It’s 200 pages. You need to brief the owner on the key findings and recommended next steps. You could spend half a day reading through it, or you could upload it and ask for a one-page executive summary.

This isn’t a hypothetical. I’ve watched architects do exactly this and get a solid summary in under five minutes. It’s not perfect. You’ll want to verify the key points against the original document. But it gets you 80% of the way there in a fraction of the time, and it catches findings you might have skimmed past on page 147.

3. Cross-checking your project against code

You’re reviewing a mixed-use project. Ground floor retail, five stories of residential above, a parking structure that connects to both. You’ve got overlapping occupancy types, multiple construction type considerations, and three different sets of egress requirements that all need to play nice together. It’s the kind of project where something slips through the cracks.

Feed AI your key design parameters, things like occupancy type, construction type, building height, and area per floor, along with the applicable IBC sections. Ask it to flag potential compliance gaps.

This will not replace your own code review. That’s not the point. The point is that it catches things you might miss on a first pass, especially on complex projects where multiple code sections interact. Think of it as a second set of eyes that never gets tired and reads very fast.

4. Generating a scope narrative for a fee proposal

You know how this usually goes. You sit down to write a scope narrative, pull up a similar proposal from two years ago, start editing it to fit the new project, and an hour and a half later, you’re still adjusting language and making sure the deliverables line up with the RFP. It’s not hard work. It’s just slow.

Take the RFP requirements, describe how your firm typically approaches this project type, and ask AI to draft a scope narrative. You’ll get a structured first draft that includes the right language, the right level of detail, and the right deliverables for the project type.

You’re not going to send it as is. But you’re starting at 70% instead of zero, and what used to take 90 minutes now takes 20. The time you save on the first draft is time you can spend refining the parts that actually need your expertise.

5. Generating a pre-meeting brief in five minutes

You have a project meeting in an hour. You should reread the minutes from the last three meetings to refresh your memory of where things stand. You’re not going to do that. Nobody does.

Instead, feed AI your last three sets of meeting minutes and ask it to flag unresolved items and open questions. In five minutes, you’ll walk into that meeting knowing exactly what still needs to be addressed. It’s a small thing, but the architects I know who do this say it’s changed how prepared they feel.

These are just five out of twenty-five.

I put together a comprehensive guide with 25 specific, practical use cases of how architects are using AI right now. Not concept renderings. Not generative design experiments. The real, daily work of running projects and managing a practice. Click here to get the free guide. 

Each one is tagged by effort level, so you know which ones you can try in the next five minutes and which ones take a bit more setup. Some of them will save you 30 minutes a week. A few of them could save you hours.