Real Value Comes From How You Synthesize Ideas: Interview with Stantec’s Arturo Vasquez

I recently went one-on-one with Arturo Vasquez, Design Principal and Senior Architect at Stantec.

Adam: Where is AI actually delivering real value in architectural practice today, and where is it not?

Arturo: Well, that’s a really good question, because it really, I would assume you’ll get different perspectives from different people, but I have a very targeted approach to AI and have seen the transformation of our practice from the early, early days of the infusion of the internet and some of the changes that happened. And this is really another transformative opportunity, and it really changes with the perspective of who you talk to, particularly in our profession.

I can tell you, from my perspective, that it is fundamental to adapt to what AI can offer if you have the right framework for how we can improve delivery, design, management, and all those things that drive our business. What we do with it here in Miami, and what we do with AI across the company, is a little bit different because the company’s perspective, with 30,000 people that we have around the world, technically, and then what we do here in Miami fuels a little bit of a difference in the approach and how we can be effective with the use of it. And that’s because at a larger scale, with a company, it’s difficult to wrap your head around the benefits and dynamics of how to use that, and the liabilities and certain things that people are wrestling with. But from my perspective, it’s a highly effective tool and technology that you can use, and we can apply, and I can give you some examples, yeah, of how we use it. We use it every day.

Rapidly evolving, and in my view really becoming a lot more minimal in what aspects of the technology, or the different softwares and the different abilities, can do to deliver and help the business. So whereas you see millions of applications, they don’t work. They don’t work for what we need to do in architecture. They may work for other businesses, but they don’t work particularly with very targeted elements of management, design, and delivery in the architecture field, in my opinion. And that’s where it’s not. Where it is is where you have done the research and done what I call the proof of concept to look at which of those applications have an effective result in improving the ability for you to deliver what you do for the business and for the practice, if that makes sense.

Adam: It does make sense. Can you give some more specific examples?

Arturo: Well, what I see, the part that is most sort of provocative in what you see, and the availability of a whole lot of tools that have come into the practice, is in the idea that technology is going to help ideate or iterate design. And that’s a huge, huge controversial discussion. Does the machine, by looking at general data, give you a better ability to design, or does it get in the way? There is no gray area. You believe, depending on what your role is in a company, that we should be using technology to help us design, or you don’t.

And I’m not in the camp of thinking that the machine is a better thinker and an ideator than what’s in my head. What it does do is it helps you create those images or those ideas and expedite the process by which you can deliver visuals or concepts in a very provocative way than what we were doing before. And I’ll give you a simple example. I’ve been doing this for 35, 40 years, and I draw. I have basically built an ability to generate concepts that take 20, 30 years to develop, because you have to absorb all of that, synthesize that in your brain, and then rapidly be able to deliver anything that a client may be looking for, regardless of the project type. And that comes from synthesizing what you’ve learned through the years that you’ve been an architect.

But it would take me a week to do a drawing, a particular drawing that’s difficult, like a technical drawing. I can do that in a day. What would take me 30 hours takes me 30 minutes. What would take me 30 minutes takes me three minutes. Because through the use of technology, we have softwares that allow me to much more quickly translate what’s in my head, in a synthesized way, into a visual, whatever that visual is. And that’s a huge game changer, because now I can develop even more complex thoughts and ideas and concepts and put that on paper or on the computer in a way that I can show it to a client, where I couldn’t do that before because it just takes too much time.

The computer can, by itself, synthesize information unless you guide it in a way that that information is what you’re looking for as an output, if that makes sense. Because if you just take any of the softwares that are supposed to be able to help you create an idea and say, I want a modernist building, it’s going to give you whatever the dataset is for that software, and it’s built into a result. Well, that’s not where the evolution of architecture is going to go. I want a building that looks like the Parthenon. Okay, well, it’s going to go look at information that is in the dataset of that software, because it’s not an open ended system unless you work with one.

But it doesn’t help you evolve that idea into something that wasn’t there before. It’s really just grabbing information or data and amalgamating that data to create what I believe is not necessarily a picture of an evolution of a thought. And only the brain can do that. It’s the same thing as playing chess. To use an analogy, Bobby Fischer had a brain, but he was creative in the way that he thought about chess, whereas when he was playing the Russians, who have a different way of thinking, they would do everything in a metric. And sometimes you come up with a move or a concept or an idea that has no foundation on anything. So how does a computer do that? How does a computer engage a thought that becomes an innovation or something that hasn’t been done before? It can’t happen by itself. It has to be guided in a way that doesn’t exist at the moment because of the way that the technology is built.

So the big specter of AI is, you know, it’s going to replace us. Well, it will if you aren’t a person in the profession or in any business that learns how to work with what is available in a way that makes you more valuable as a tool. You can bury your head in the sand and think it’s going to go away. It’s not. It has already changed everything we do, and we’ve already implemented, in one year’s worth of time, what would have taken me most of my career to wait for until the technology was available for us to do what we’re doing now.

So I’m perfectly happy, because now I can do things that I couldn’t do before. I’ll give you an example. I can take any sketch that I do, and a lot of my colleagues are also very good at doing what they do, and we grew up in the practice the same way. And I can take this image, and I can show you the backside of the image without doing anything, because AI will interpret what it’s seeing, run through the dataset, and give you the definition of an angled view that you didn’t draw before. Well, I would have to draw that angle view. I would have to draw that rotation, just to put it in words.

If I start with an image, like an X-ray of what happens to your body or whatever, well, you’re only going to see that image. But look at what’s happened with the healthcare profession. Now you’re getting three-dimensional views of everything, so the diagnostic for whatever is impacting your health is much easier to get to the effective medicine or protocol for that to happen. The same thing with architecture. If you use it in that way, then you’re able to see what you couldn’t see before, because you would have to either create a model of it on the computer, and it was a static thing, or draw all the faces of that image that you’re trying to convey to the client.

What does it look like from the front? What does it look like from the side? What does it look like from the top, bottom, left, and right? I can do all of that in one. I can actually dynamically show you what would normally take you multiple images, because they’re all fixed images. That is a transformation in what we do, and there are two or three specific softwares that do that. I call it cinematic progression.

We didn’t have that before. Movie makers make a storyboard. The director makes a storyboard. That storyboard is a series of little vignettes. Well, now the storyboard is almost a three-dimensional understanding of the scene. It’s the same thing for architecture. Now you can get a client to look at the building from the outside, from the inside, from the top, from the bottom, whatever you want to see, I can show you.

Which means that you have to be a lot more sophisticated in your head to visualize the entire sense of the design concept that you are conveying to a client, and you add huge value if you’re able to have the client understand what you’re seeing and what you want to give them for what they’re hiring you to do. So it’s a big different world in how we have to think about design and architecture with the technology that’s available that is giving us that opportunity to do that.

You can take a map of the world, and if, like me, you’re a fan of maps, and you see maps of the world, you know they’re fixed images because they didn’t know where the limit was. And take that globe behind you and spin it. Technology can do that for you because it’s interpreting the information in order for that globe to spin relative to a map. Now you can actually take the map and spin it, and it will create the elements that it doesn’t have, because it’s reading the elements that it has. It’s a huge, huge transformation in the way we convey information, we convey images, design, and execution of those thoughts.

Adam: What tools are most important for anyone in architecture to understand and leverage?

Arturo: Well, after looking at a number of softwares out there and protocols, and there’s a lot of people putting in, you know, now we can do X and we can do more of X that wasn’t there before, but it never really reaches the goal that it needs to reach, what we’ve figured out is we use, for a lot of the narrative work, we use Copilot for a lot of the narrative work, simply because it’s built into almost everything you do. So, you know, I can Copilot this conversation, and it’s giving me a readout of what the computer already understands about the way I think, that’s the way Copilot works to some extent, we can write a code sheet, an analysis on a project, by looking at all the projects that we’ve done where a similar set of situations have happened on a particular site, to analyze through Copilot a potential understanding of the zoning which would impact that project without looking at the zoning, I mean I would spend the whole week just looking at the zoning for that site to understand what it does, but if you use Copilot in the way that we use it, it will analyze every project that our company has done that has similar parameters and it gives you a readout and says okay well this may apply, this may apply, this may apply, this may apply, and you can do that for almost anything, you can do that for emails, you can do that for responding to proposals, we call it request for proposal, we respond to RFPs, well I can ask Copilot, or my colleagues can ask it, give me a readout on what we would need to answer for similar requests for proposals for this type of project and it will give you a rundown based on the dataset that you tell it to read from, which is all in the computer databases that you use for the whole company, so it will go through almost every project that anyone in the firm has available and redact an example, and of course it’s not going to be perfect, but it’s going to save you a lot of time versus doing that yourself and going line item by line item trying to figure that out.

So it’s helped us on the narratives, on how we respond to proposals, on looking at the pros and cons or what we need to focus on for this type of response, we had a recent situation where we submitted a proposal, we won, and then the municipality kicked it out, and then they re engineered what they wanted for the same proposal but then re awarded it, so our client is saying well how do we resubmit, my colleague just asked AI through Copilot what in this document is different than the original document that we responded to and what would we have to change in our proposal and give us a comparison, and it told us exactly this is what you submitted that responded to this proposal, now they’ve reworded the proposal, now this is what you need to submit, and it cut everything to ten percent of what we really needed to do, as opposed to a human analyzing that, because it takes time and you would have to be really good at comparing what you submitted, reading everything, and then trying to interpret what the new submittal was, so somebody changed the parameter, let the computer figure out what that parameter change is and how you would adapt your response so you don’t waste time and money answering that.

Fast forward, and that’s all the narrative stuff, contracts, liabilities, how we set up a response for a proposal, even some of the narratives around how we write a cover letter, I can write a cover letter in twenty minutes, but if I use previous projects where we approached a similar response it will give me my own thoughts back in a structured way and then it becomes better, but I have to start with something, if I just let it run on its own and I’m not guiding it, it’s just going to grab whatever it grabs, because you’re not telling it take my cover letter, look at what I’ve said before, and see what’s missing, and we do it like this really fast.

On the design side, we’ve narrowed the focus to softwares that already belong to the company so we don’t have to go buy another software and another software and another software, our fundamental drawing tool is Revit, which almost every architect uses, but before we use Revit we’ve created an analytic tool based on a Dynamo script, which is a way that we create what I call the metrics for a project, like I’m going to design your house, okay tell me what you want, I want a two bedroom house, I want a large living room, I want a three car garage, that’s called the metrics, it gets more complicated with healthcare or mixed use projects, but the idea is the same, we developed a tool that allows those metrics to be manipulated based on our knowledge of translating that data into form, meaning a visual form, you wanted this, here is what that looks like, and here is all the criteria and the result based on that lot, that campus, or that parcel where we’re working.

Then we use a software called D5, which is a very advanced cinematic visualization tool that gives you lighting, trees, people, materials, I can design the entire environment so I can walk you through the image instead of hiring a rendering consultant to give you one static image that costs a lot of money, we’re not designing fixed images anymore, we’re designing entire environments that can be cinematically visualized and respond to the metrics that make a project viable, and then on the back end instead of buying software for context we use tools like Gemini to generate that context, but you have to know how to prompt it, the language matters, the trial and error matters, and the prompts we’ve developed matter, so what happens is the Revit output from the Dynamo script feeds into D5 and then the AI tools enhance that into a full cinematic expression that would not be possible otherwise.

And that’s huge value, because before you needed four people, someone for analytics, someone who understands zoning, a designer, and someone who knows the software, now it’s basically two of us, me and someone younger who is a digital native and brilliant with technology, every Monday we sit down, I say where I want to go, and he figures out how to get us there with the technology, either we adapt what we have, we find something new, or we drop the idea, and it completely expands your ability to do what we do as architects without needing the same scale of resources.

Adam: When architects use AI tools, whether they’re generative design, performance analysis, or documentation, where do those outputs break down and create risk?

Arturo: You just mentioned four terms that are spot on with what we’ve been talking about. That language you just used already shows an advanced understanding of where we are with that. Where I think it falls apart, and we just had this discussion today, is when different softwares don’t align, and that has always been a problem. So, for example, buildings are very simple and normative, meaning they are straight lines. A building is a bunch of straight lines. But what if the building is not a bunch of straight lines? There are architects out there like Zaha Hadid, MAD Architects, Bjarke Ingels, doing what we call complex geometry. In order for you to become an architect, you had to pass one class, descriptive geometry. My father told me that if you want to be an architect, you need to pass descriptive geometry, and that was actually true.

Descriptive geometry allows you to draw in three dimensions what you don’t see. You take a cube, and you have to be able to draw, through an understanding of descriptive geometry, what the back of that cube looks like. In medicine, if you want to be a doctor, you have to pass anatomy. My brother was a doctor. If you can’t pass anatomy, forget it. Same idea. Now, think about a circle or an oval, or if I take a cube and I want to twist it. You can’t do that with any software. You need software that was invented to design complex geometry, because there is a changing curvature. There is a software called Rhino that is very good at it. A lot of younger architects learned Rhino early, so when we interview someone, and they list software, we already understand what they’re capable of doing. The problem is how do you take those visualization and modeling softwares and seamlessly bring them into drawings. Because at the end of the day, in order to build anything, we still have to take three dimensional information and put it into two-dimensional form. That’s how you build a car. That’s how you make a suit. You take something three-dimensional and you template it.

We do the same thing in architecture. Very few softwares are able to combine both things well. We’re not there yet. When AutoCAD was around, it was two-dimensional. Then we moved to object-based modeling, which allowed us to work in three dimensions and then generate two-dimensional drawings. Now we can also move through those models cinematically and detect clashes, which was not possible before. So the technology helps, but the flaw is that different platforms do not merge. I don’t think there will ever be one universal system. They are owned by different companies, built differently, and used differently across the world. So there will always be multiple softwares working together, and that is where the risk is.

Adam: In what ways does human judgment matter more today because of AI, and in what ways does human judgment matter less because of AI? And what will separate the architects who use AI well from those who don’t?

Arturo: I think there’s a better word than judgment, but you’re going in the right direction. It doesn’t translate perfectly, but it’s really about having maturity and being judicious about the decisions that you make, which is not the same as being judgmental. The problem with AI is that it is judgmental. You can have very young designers that know nothing about architecture implementing sophisticated AI and coming up with a solution that is not grounded in anything that has value, because there is nothing behind it other than some kind of image that was created by the computer.

I’ll give you an example. There was a competition, and in that competition, you had teams that were very good at using technology. Let’s say you are my client, and you have a logo for your business. I present to you a solution for your new headquarters, and I say we used this software, we took your logo, and we made it into an origami, and that origami is now the facade of the building. That’s what you’re going to pay for. Do you think that solves the vision of what you’re looking for? Did you hire an architect to come up with a vision for your aspirations, or did you hire someone to manipulate a logo into a shape? I would fail that person in my class, because that person has not provided value for what they’re being paid to do.

That’s the problem. A client is asking for a synthesis of problems to be translated into physical form so they can perform their business. We just did a project for FIU where we had two completely different clients, an academic client and a healthcare client, that see the world very differently and had never worked together. My job was to give them a building that met the aspirations and requirements of both in one physical form. The computer cannot do that. It cannot understand what one institution represents, what the other represents, and synthesize that into something meaningful. It will pull from data sets and produce something, but it won’t necessarily be thoughtful or grounded.

Architecture is fundamentally about solving problems. It is programmatic, philosophical, and operational. It is not just form-making. If you were really an expert at using technology, could you and your spouse build your own house? Probably not. It takes someone who understands people, constraints, and how to synthesize those into something real. That is what separates people who use AI well. It is not about the tool. It is about the maturity of thought behind how the tool is used.

Adam: How else can architects or anyone in a leadership role within an architecture business effectively leverage AI?

Arturo: I think, if there was, for architecture specifically, the biggest disconnect is the practice and the business. We are not trained as architects in school to understand how to run a business. You learn that on your own if you want to. I ran my own practice, and it took me ten years to learn enough to where I was comfortable that I could run a business, because that was not taught to us. You don’t go to design school to learn how to run a business, but you should. My best students graduated in architecture and then went and got an MBA. An architect who has an MBA is rare, but then you understand both the creative process and how to run a business, and those two things have nothing to do with each other.

Most companies, whether it’s a small boutique design firm or a very large corporation like the one I work for now, operate in a split way. I started in a boutique design environment because I wanted to focus on design. Most boutique firms don’t survive because they don’t know how to run a business, unless they have financial backing and don’t need to worry about it. If you’re not in that situation, you have to earn a living and run a business, and they don’t teach you how to do that. In smaller firms, you learn design but not business. In larger firms, you learn business, but you may not get the same level of creative opportunity. It takes ten or fifteen years to learn both sides. If AI could bridge that gap, that would be transformative. If AI could help architects understand the business side, and help business leaders understand the creative side, we would win as an industry.

Right now, that doesn’t exist. The people in accounting and finance have a completely different mindset. They use different softwares, think differently, and operate differently. I can’t easily interpret the financial reports they produce, and they don’t fully understand the creative and client dynamics that impact those numbers. So there is no system right now that connects those two worlds. On the creative side, what I am starting to do, and I figured this out very recently, is try to understand the business model behind what we design and bring that into the way we think about projects.

For example, I can design a house and show you something beautiful using all the tools we talked about. But when you ask me how much it costs, I don’t have a system that connects that design directly to cost in real time. That doesn’t exist. I can estimate based on experience, what we call rough order of magnitude, but I can’t guarantee it. That requires a contractor who has access to real pricing data. They use completely different systems and don’t share that information. If we could combine design, metrics, and cost into one system, that would completely transform the practice. Imagine designing something and immediately knowing how much it costs, adjusting it in real time, and giving the client instant feedback.

The people who run accounting have a different mindset. In order for you to be a good accountant, your mindset is different than the creative person. So we have a very difficult industry, because we have to be creative and also run a business, or have the business mindset, and there is no real way for those two mindsets to come together right now because we use completely different softwares. I can’t even tell you how much time I would have to spend to understand some of the reports that I get, where all I see are numbers that represent financials. That’s what those people do every day. They understand those metrics immediately, but that’s not what I get paid to do. And they don’t understand why a project is losing money, because they don’t understand the client dynamics or the design decisions that impacted that outcome. So no, right now there is no software that connects the financial side of the practice with the creative side in a meaningful way. That is going to take time, if someone is even working on it.

What I am starting to do, and this is very recent, is try to better understand the business model behind what we design and bring that thinking into the design process. For example, we went to a meeting with a development partner, and I realized that the more I understand how a project is financed and structured, the more valuable I become in shaping the design. But we still don’t have a tool that allows me to run that business model alongside the design in real time. I can design something and show you a beautiful building, but when you ask how much it costs, I don’t have a system that gives me that answer instantly. That doesn’t exist. It only exists in my head based on experience. I can give you a rough order of magnitude, but I cannot guarantee it, and I’m not going to sign a contract on that basis.

To get real numbers, I have to go to a general contractor, because they are the ones calling suppliers, pricing materials, and understanding the real costs. They use completely different systems, and that information is not shared in a way that integrates with our design tools. If we could combine those worlds, design, metrics, and cost, into one system, it would transform the industry. Imagine designing something and immediately knowing how much it costs, adjusting it in real time, and giving the client immediate feedback. That would change everything. But right now, we missed that completely. When we submit proposals, we are often estimating costs without full certainty, and that can cause us to lose projects because we either overestimate or underestimate. There is no tool today that allows us to do true target value design in real time. We rely on experience and external partners, and that gap is still very real.

Adam: What the biggest ethical considerations for architects when working with AI?

Arturo: Plagiarism. Straight out plagiarism. Intellectual property violation. AI is built on data, and we don’t always know where that data comes from. That creates real risk. There are a lot of arguments around this. Some people say it’s no different than looking at something in the real world and drawing inspiration from it. But I don’t know what dataset a tool is pulling from. Is it public? Is it proprietary? That’s the issue. In architecture, if something is visible in the public realm, you can interpret it and draw it, and that is not considered intellectual property violation. But AI is different, because it may be pulling from closed datasets that you don’t control or understand.

So how do we deal with that? The best approach is to build your own dataset. Take your firm’s work, your own projects, and let AI learn from that. Then you are evolving your own language instead of borrowing from unknown sources. That’s what the best firms are starting to do. They are creating internal datasets so that the AI is working from their own intellectual property. Because if you just tell AI to give you something that looks like another architect’s work, what value are you adding? You’re just copying in a different way. In our firm, we have strict rules. We cannot use AI-generated images unless they are based on concepts that we created. We cannot just let the machine generate something and present it as our work. That is the biggest ethical issue right now.

Adam: Given the proliferation of AI in architecture, what do you believe will separate the best firms from everyone else because of AI?

Arturo: Speed and maturity of ideas. AI allows you to generate more ideas faster, but the real value comes from how you synthesize those ideas. When you’re young, you might come up with five ideas. As you gain experience, you can generate many more. AI accelerates that process. But the client does not want one hundred ideas. The client wants you to tell them which ideas matter and why. So the firms that succeed will be the ones that use AI to enhance their thinking, not replace it. They will be able to take more information, process it faster, and arrive at better decisions more quickly. It allows you to work at a level of complexity and speed that was not possible before, but the human element, the ability to synthesize, to decide, to guide, that is still the differentiator.

Adam: Is there anything else you would like to share?

Arturo: I think if we have this conversation a year from now, it won’t be that much different. The tools will improve, but the fundamental structure of how we work will not change dramatically. The business side will still operate the way it does, and the creative side will still operate the way it does. Right now, I cannot manage my team through a system that tells me in real time how much time we are spending or how profitable a project is without going through accounting systems that I don’t fully understand. When those systems start to connect, when design, business, and execution are integrated, that is when things will really change. But today, that connection does not exist.