Anyone can prompt an image that looks complete. It is hollow inside.

Hollow on the inside

Lam’s answer is one sentence long and it sets up everything that follows. “Clients (or anyone) can prompt images that look complete but are actually hollow inside, and sometimes it’s difficult to argue against or prove a computer wrong.” An architect’s traditional advantage was being the only person in the room who could show you what the thing would look like. That advantage is gone, and what it leaves behind is a conclusion he states flatly: “In this time I feel trust is everything.”

The commercial effect is not the one you would expect. Fewer people call, and the ones who do are better. “The availability and accessibility of AI images and outcomes means that people who talk to us are actually those who value our design input, and know that nuances and cohesion count for something. It’s a smaller pool of potential clients now, but there’s more trust that we are able to deliver something human.”

What that remaining group is buying is not a picture. “The emotions matter — being surprised by something, seeing beauty in the imperfect, and knowing that craft and discovery is much more interesting than outcomes shaped by machines.”

And one thing has quietly gone, which is the loss he seems to feel most. “I feel like there’s less open-minded wonder about what is possible, because AI has at least put some options or approaches in their heads.” People now arrive with an answer rather than a question.

On whether anyone is willing to spend more than two years ago, he gives the mixed answer rather than the clean one. Right now the local economy is under cost-of-living pressure, families are avoiding unnecessary expenditure, construction costs are still high, and “it seems bang-for-buck is a critical consideration now.” Zoom out, though, and the line goes the other way: as Brisbane has grown as a city, “the appreciation for design (and the quality of outcomes) has really come up over the last decade.”

What a client brings tells you what they value

About half arrive with a Pinterest board. The rest come with a blank sheet or an open mind. Most are working with an architect for the first time and are “a bit unsure in terms of how prepared they should be”, and he does not think any of that has changed in two years.

What has not turned up, despite everything above, is the thing you would assume. “We’ve yet to see an AI generated image that is actually the ideal solution for a given project or site.” Renovation work is where it falls over hardest: nothing he has been shown “takes into account structure and constructability, and therefore cost.” The picture does not know what holding the roof up costs.

Clients, for their part, are careful about it. “Most clients are respectful — they may say something like ‘we’ve AI’d something, but we don’t want to constrain your thinking.’” And the studio has a settled way of receiving whatever lands on the table, which applies equally to a hand sketch, a Pinterest pin or a generated render: treat it as informative, “in the sense that it reveals something about what they value.” Then the honest caveat, said up front. They will give it exactly the respect they give their own ideas, “in other words, it may get discarded, absorbed or reshaped as we step through the Concept.”

The site offers a downloadable Client Starter Pack, and we asked where it came from. The answer is a small lesson in how a marketing asset outlives the positioning that produced it. The old branding was built on being open and transparent, and it attracted clients who loved being involved in the process, so the studio made them something to prepare with. He is not precious about how it stands today: “The one that’s on there is pretty work-in-progress, to be fair. I’m not sure that our revamped branding is resonating with the same people as before.”

The drawing is not the point of the drawing

Where it has reached is scheduling products: collating information and understanding price points, which he calls a genuine strength of the data models. Where it has not reached is almost everything else, including the obvious one, rendering. The reason is the most interesting thing in the interview, because it is not about quality.

“We’re always iterating as we step through different components of the drawing package. A Reflected Ceiling Plan, for example, is not just about producing the drawing but a chance to design lighting and how we see the AC/mechanical ventilation systems working.” The drawing is where the designing happens. Automating it would produce the artefact and skip the thinking. He is explicit that this is a trade and not a verdict: “AI could possibly improve the efficiency of these parts of the workflow, but it would be a loss to surrender the opportunity to refine our designs as we step through.”

The cost side is a single sentence and a single incident. “Definitely makes information more accessible, but it presents truth and misinformation with the same confidence.” Then the example: an AI gave them the wrong zoning overlay for a site, and a graduate did not catch it.

The experiment that failed was not theirs. “We’ve had clients AI their own alternative colour schemes, for example, and it created an insane amount of work for us to coordinate samples on their behalf, just so we can prove to them that the AI scheme was not a good fit for their project.” Someone generates an image in seconds; the studio spends days with physical samples demonstrating why it does not work.

Which leads to the one detail in this series that runs backwards. Elsewhere in these interviews a client tried to write a no-AI clause into a contract. Here it is the other way around: “in our Client Architect Agreement we specifically ask them not to use AI over our work.”

Paying big-practice prices in a small studio

The stack is deliberately short: Google Docs and Sheets, Adobe Creative Suite, ArchiCAD and Twinmotion. Asked whether anything newer has actually got better than the old CAD giants, he does not hedge. “Definitely, 100% ripe for disruption — we can only pray and hope. It’s ridiculous that a small studio like us is paying for big-practice tools that are of no use to us.”

So the bill is going the wrong way, and not because of AI. ArchiCAD is moving customers onto subscription, which for them means a significant overall increase. Their response is the one every small studio recognises: buy the final year of perpetual licensing “and stay on this version until it becomes obsolete.” A deliberately dated tool, chosen on price.

Two things got cancelled recently, both for reasons worth stealing. Microsoft Office went to Google Docs and Sheets: “after some complaining the team got used to everything pretty quickly. Just tried to shave costs wherever I could.” And Programa, a scheduling tool built for interior designers and architects, went back to a Google Sheets template. The reason is not about the software at all. “Programa made pretty Schedules but the technical information got lost and we couldn’t trust Builders to consistently access something that was mainly online.” A tool only works if the people downstream will actually open it.

What he still does by hand is the part nobody enjoys. “Everything related to invoicing and budgeting. Xero doesn’t automate payment receipts, and we’re still inputting data back into a Google Sheet to understand our numbers over a given month.”

Two cost checks before anyone commits

The sequence is unusually explicit for a small practice, and two of its steps are there specifically to stop a project from drifting away from its budget. It opens with a detailed briefing meeting where the team get to know the client “and what makes them tick”, then a Return Brief that tests whether the expectation and the early budget are realistic. Concepts follow, sometimes more than one per brief. Then an independent party, usually a builder, provides a cost estimate “to cross-check our understanding”.

Design Development is the bulk of the work, with consultants brought in to confirm scope, and a second cost estimate lands at that point. Final documentation and interior details follow, the cost of works is finalised, and the studio stays around during construction “to help the project come to life”.

Everhour sits underneath that, and Lam frames it as a tension rather than a tool, which is a more honest description than most. “How does knowing that I’m running over on fees help me to finish a house well?” Billable hours are a commercial reality and a poor KPI for creative work at the same time, and what he says the tracking buys him is the ability to hold both: “it offers the right amount of information and accountability without being a burden to the team.” They are still predominantly interested in making work they would want to live in themselves. “Without this information we’d be navigating blind.” He is careful to point the same lens at himself. The studio is always looking at how it can improve internally, “there’s always room to be more efficient, less double-handling, learning lessons along the way”, and then the half most owners leave out: “but also we’re always learning as managers (and that includes the leaders setting realistic fees for a project).”

There is no project management system behind it, and he was asked bluntly whether there should be. “We’ve tried using Monday and I’ve heard of other Architects using AI to create project dashboards, but to be honest I don’t think we need to overcomplicate things.” Project leaders allocate subtasks to the team in a Google Chat space, and that is the whole of it.

On whether projects land on time, he gives the answer the question deserves. “We’re still usually running behind time, especially towards the end parts of the project.” The studio is trying to get more efficient as it grows, and what the time data buys is not a cure but a diagnosis: “having time insight allows us to develop processes to help address the pain points.”

Which clients cost you, and which ones do not

Fees are structured by stage rather than set once. Concept is a flat fee, which gives the client confidence at the first commitment. From Design Development onwards it becomes a percentage of build cost, so the fee moves with the cost of works as that evolves. The construction stage is a monthly retainer.

When a project takes more hours than the fee assumed, the answer is one word and an exclamation mark: “We do!” Past projects feed the next proposal, but he is clear about the limit of that in the current market: “in a down-market it’s hard to keep our fees up too.”

The thing that hurts is not the overrun, it is the gap. “Cash flow is everything, and it really hurts us to have a hole in our program.” A stalled approval or a client who goes quiet does not cost a fee, it costs a month of the studio’s capacity.

Tracked time has changed how they price, and the finding is specific enough to be useful to anyone running a studio. “One thing we’ve seen, as a practice, is how much time gets absorbed in working for clients who are emotionally invested in their ‘forever home’. With good tracking we can see that projects for Developer clients can be priced much more competitively.” The same tracking lets them go the other way for people they want to help: being transparent and accountable with their hours is what makes a capped maximum fee possible, which they offer “to select clients, usually in the non-profit or church sectors.”

And the number he now watches hardest is the one a lot of owners avoid looking at. Overall staff efficiency — wages as a proportion of income — “has always been our achilles heel”. Two years went into process and marketing. “This year the main goal is to get the business fighting fit as we become an established practice in the marketplace.” And the metric is only legible because the hours are there: “time data is obviously critical to our understanding of this issue.”

“Few grads actually make a case for what they would add”

The studio has stayed small, with more experienced staff in it than before. Graduates are no longer being hired, and the reason is unusually candid: “We used to be grad-heavy, and to be honest I think I’ve done my time and given a few their big break in their industry.”

What he thinks someone starting out should learn has nothing to do with software. “Authenticity is everything. I think people are only moderately impressed by slick visuals — that demonstrates some level of competency, but character is the most important thing.” And then the observation that should be read by anyone about to send an application: “Few grads who apply to work with us actually make a case for what they would add to the office.”

Asked whether AI has changed what he expects from a good employee, he answers in one word. “No.”

The last question got the longest answer in the interview, and it is the only one where he argues for something rather than describes it. He hopes people will get tired of AI-derived things and become naturally suspicious of images that look slick and too perfect. Then, “instead of judging a work of architecture or design through a 2-second glance, perhaps people will be more interested in the story and reasoning behind what they see. And when they do so, they’ll realise that visual beauty is only one tiny aspect of something much deeper.”

He reaches for music to finish it. Every piece of art “is an expression of something bigger, and the lived experience of the person who created it matters. Conversely, how hollow it would be if the creator had never felt anything of the emotions or ideas they’re trying to convey.” What he wants is room left for architects with something to say: “as we shape good spaces there’ll be a tangible difference, because our work is founded on deeper foundations, and it’s not just all talk. The things that we fight for makes a difference in the lives of the people who inhabit our buildings. This is what we long for anyway!”

Seven short ones

One industry cliché you don’t believe any more?

That we need to create a studio around a single Starchitect and their persona.

Which piece of common advice in your industry is wrong?

That we would succeed if we just follow the footsteps of those who have gone before us. Things have changed too much between generations for this to be true.

One thing every practice should stop doing?

Adding noise to the world without having something good to say.

One thing you wish you had started doing earlier?

Taking self-marketing and self-promotion seriously; I was too embarrassed to do this.

Something you would never automate?

The full Concept process — that’s the best part of this whole profession.

One tool you couldn’t run the studio without?

Everhour?

A prediction two years out you’d bet money on?

The barrier of entry will keep getting lowered, but it’ll be harder and harder to build something that lasts.