Practice Rebuilt: AI's real gift isn't speed, it's making you question how you work


Hello Reader,

The story we keep telling about AI in practice is that it makes the work faster. Draw it faster, document it faster, and get the deliverable out the door faster. That might be the least interesting thing it does.

The more useful thing, and the one almost no firm is built to use, is that it makes us question something we stopped noticing a long time ago: why are we still working the way we always have? It's the question I kept circling in this week's bonus episode, a conversation with Ross Wagner of Pirros and Xander Ellenbogen of Clark Nexsen about giving a firm a searchable memory of everything it has ever drawn.

/// Listen Here:
Never Draw It Twice, Giving Design Firms a Working Memory

The harder truth is that most firms aren't especially good at optimizing how they work. We repeat: the same process, the same setup, the same steps, largely because that's how it's always been done. We rarely stop to refine it, and we almost never step back to ask whether the whole approach still fits, or whether we're simply running the process we inherited. Most of how a firm works was never actually chosen. It accumulated.

AI is useful precisely because, used as more than a grand search engine, it interrupts that inertia. When a tool can produce in seconds what your firm has always produced by hand, the way you've always worked stops looking inevitable, and you get to ask whether it was ever the best way or just the only one you knew. Ross put a version of this to me plainly: Pirros isn't there to flatten creativity into copy and paste, it's there to stop people from spending hours hunting for something they were only ever trying to learn from. Take the hunting away, and what's left is the learning it was in service of.

That learning is the whole point, and it's what the next generation of architects actually needs from us. The worry that automation hollows out their path has it backward. When you let go of the work that doesn't serve you, what's left is the work that was always the real education: the reason a decision was made the way it was, the call no library can make for you.

So the question isn't how do we do this faster. It's why we're still working the way we always have.

Where in your firm are you doing something the way it's always been done, without anyone ever deciding it was the right way?

Keep learning and growing,

Evelyn M. Lee, FAIA | NOMA

Founder, Practice of Architecture
Host, Practice Disrupted & Fractional COO


If this week's episode landed for you, it's worth knowing that they're building the thing Ross and I were talking about: a searchable memory of every detail your firm has ever drawn, so the next person never starts from a blank sheet. If you're rethinking what your firm should stop redrawing, see how it works.


/// practice disrupted /// - latest episode

Ross Wagner of Pirros and Xander Ellenbogen of Clark Nexsen on turning decades of a firm's buried details into a searchable, shared knowledge base, and what that does to change management and the apprenticeship model.


/// the signal /// - news you may have missed

Three signals I wouldn't skip this week, and what they say about where the value is really going.

Common Edge: Chad Reineke on how AI makes design look easy, and how that quietly re-prices your expertise. The firms that will struggle are the ones still charging for the output rather than for the judgment behind it.
read here

Autodesk is buying MaintainX for $ 3.6 billion, the largest deal in its history, and it's a move into how buildings run, not how they're designed. Read it as a signal about where the value sits over a building's whole life.
read here

EntreArchitect on the architect-as-developer model, where owning a piece of what you design beats billing for it by the hour. One honest answer to the question of what to do when the drawing stops paying.
read here

/// the handoff /// - something you can do on Monday

Pick one thing your firm does every week out of habit. Before you write it down or hand it off, run it through four questions, in order:

  • Does this still need to happen at all, or are we doing it because we always have?
  • If it does, does it need a person, or can a tool do it now?
  • If it needs a person, does it need to be you?
  • Whatever survives all three: is that where your time is actually going?

Most firms only ever ask whether it needs to be them. The real gift of AI is that it lets you start at the first question instead of the third.


/// up and coming ///

Season 13 of Practice Disrupted opens Thursday, July 30. Until then, the feed belongs to a wonderful word from our partners, starting with the Trimble 0-60 bonus episode.

Evelyn Lee

Read more from Evelyn Lee

Hello Reader, NCARB just released the 2026 edition of NCARB by the Numbers. It is the fifteenth. For fifteen years, the profession has been able to see the path to licensure plainly: too long, too rigid, quietly losing people we should have kept. For most of those years we argued about it and changed little. What is different now is that the numbers are finally reshaping the path itself, into something more flexible and modular, built around where the data says people actually fall out. This...

Hello Reader, My latest episode of Practice Disrupted is about AI closing the gap between a sketch and a built thing. I sat down with the team behind Trimble's 0-60 program and Formas.AI, and we discussed how quickly those tools are moving. /// Listen Here: Closing the gap between sketch and build, with Trimble 0-60 and Formas.AI. But the tools were never the part that kept me up. The harder question is what happens to the practice around them. Practice of Architecture is a business in...

2026 Hello Reader, There are small moments with my kids I would give a lot to keep, the offhand thing said on an ordinary Tuesday that you only realize later you wanted to hold onto. Most of us try to save those by typing out what we remember, which means we mostly save what was already top of mind. A real conversation does more: it asks the next question and pulls out what you didn't know you knew. I saw a version of this last year, a demo at a Conference of Mayors meeting where AI gathered...