Hours Down, Value Up: SEG CEO Ali Fakih on Building AIAEC

PSMJ Resources, Inc.
Posted on: 11/16/26
Written by: PSMJ Resources, Inc.

This blog content comes from an interview conducted on October 13, 2026. For more information or context, watch the full episode where you get your podcasts.

“We have a client that literally hired us to do a due diligence report that usually takes our guys about 200 to 220 hours. Now we’re able to run the first draft in six minutes.”
Ali Fakih, Founder and CEO, SEG and AIAEC

Most engineering firm leaders worry about what AI will do to their hourly business model. Ali Fakih built a company around it. Fakih founded SEG, a planning, entitlement, and civil engineering firm headquartered in Phoenix, in February 2009, at what he calls the beginning of the worst real estate crash ever. Almost three years ago, he started AIAEC to bring AI to the industry. On this episode of the AEC GrowthCast, host Greg Hart talks with him from both sides of that desk.

 

1. Let the experts shape the AI

“Three years ago, when AI started taking more space in all of our lives, I was one of the people who believed that subject matter experts and professionals in our field have to be the ones figuring out how we incorporate AI into our workflow and how we get it to work with us and for us, to address similar challenges and similar pain that our professionals have. And this is exactly what we did.”

His starting point was the pain he saw at SEG every day, and permitting topped the list: “Everyone complained about permitting. Municipalities complained about permitting. Politicians complained about permitting. Developers, as professionals, complained about permitting. So there is a huge pain in the permitting process.”

He also found that cities want the same outcome: “What we learned throughout the process, especially looking at it more from the municipal side, is that their goal aligns exactly with our goal. The perception is that the municipality is dragging the permitting. They would love to have the quickest permitting, but the challenge is way beyond what they’re controlling.”

And the work is mostly administrative: “Before you even have a solid professional making the last call on a plan, 70% of it goes through administration, which is insane.”

2. Get comfortable being uncomfortable

“In July 2023, I would be sitting in a room and they’re talking about model training, AI, computer vision, how you get into the right model and guardrails, and honestly, I had no idea what it was. It was way beyond my head. At one point, you start figuring out the business side and what your added value is. My added value was more as a product guy, someone who’s seen a pain and envisions how he wants to fix the pain.”

He’s open about how much he didn’t know: “I’m a big believer that you have to be comfortable to be uncomfortable a little bit. You have to, and if you don’t, you’re stuck. When I hired our first engineer, I had no clue, truly no clue. I built the strategy in the first six to eight months while I’m learning through it. Today I know a lot more about the technicals than I knew three years ago.”

3. Start with the everyday work, and show each audience its own win

His municipal pilot began small on purpose: “They just gave us six permits to start with. We all see those huge permits, the big mixed-use projects. But by city, how many of those projects are there compared to the majority, from single-family homes and pool permits and tenant improvements? That’s where the majority is. You have to focus on their pain today, which is the majority of the amount of work before it comes to them. It’s like, ‘I want to do your most important project through AI.’ No. Let’s start, get your trust where you need to, and then we can go from there.”

Each audience wants something different from the same tool. Developers: “How does this help us in this competitive market to tie up a property earlier and finish entitlement earlier? What’s the dollar value to it? How can you help us find a property that we could do what we want to do on it?”

Brokers: “What they love about it is that early on, when they come across a property, they don’t need to have a call with a land use attorney, an architect, or an engineer, sometimes asking a favor, sometimes paying for it. They just want to know what could be done here. What’s the height restriction? What type of use could be here, so they can know who to send this property to and start going after it.”

And architects. Fakih relays how one user described a client call: “We’ll be with a potential client on a phone call, just talking about potential properties, and we put in the address. And while we’re talking, in 30 seconds, we’re able to pull in zoning information. What’s the parking requirement? Is he able to do what he needs to do? Just sourcing that information within 30 seconds positions us as knowing the city so well and knowing the project so well that 90% of the time we’re closing the deal just based on this little information.”

4. Give professionals time back, and keep a human in the loop

“I have not met a single professional engineer, planner, or architect who’d sit down with you and say, ‘I’m so bored. I wish I had a little bit more to do.’ I have not met a single person, especially in the last eight, nine years.”

That’s why the pitch works: “AI will not replace a human on what you do. Even if you forget about the license and whatnot, you still have some touch that we humans have to have in our field. There’s a lot of liability, there’s public health, there’s tons of creativity that occurs. Our part is so important, but there are a lot of things we do today that we could easily reduce, to make us more productive on the things we like and enjoy the most.”

Does a faster due diligence report cut into SEG’s billable hours? Fakih says clients still come back to the firm: “I had a client call us and say, ‘I’m using the tool. The due diligence you charge us for, it’s making life a lot easier.’ But they still use us on the other side. There’s a lot of analysis and work that has to be done early on, and you always want to have a human making the final decision. Even if they run a due diligence, they want this touch of a professional to confirm it.”

When a developer asked how much to trust the tool, he offered an analogy: “You have two approaches here. If you hire someone fresh out of school to help you with due diligence, and they just email it to you and you’re the type of person who will just forward it without even opening it, this is not for you. If you open it and check what they did, it means you still add this second set of eyes that’s needed early on. And if you set this expectation time by time, you start building trust with your tool, and you’ll know when you don’t need to open it and just forward it.”

5. Domain-trained data beats a general-purpose chatbot

Prospects sometimes ask why they can’t just use ChatGPT or Claude. Fakih’s answer is about data: “It all goes back to trust and training, and it’s all about the data. It’s not the perfect analogy, but I’m going to use it. Say you have a nine-year-old and you give them two books. One has all the wrong ideas in the whole world and they keep reading it, or you give them one with truly the right ideas on whatever learning experience you want them to have. If the data is good, they’re going to be able to use it and build on it.”

He sometimes demonstrates the difference live: “We ask those people exactly this question: pull up the general AI and ask. Is it going to pull information from everyone? You could be lucky to get information specific to the parcel, but you may get a Reddit opinion about the zoning.”

The difference comes down to guardrails: “This goes back to how you can set up your model with the right guardrails to only focus on the items that relate to the vision of the product. This is the biggest part of it, and we’ve been spending so much time on it, even in due diligence, when you’re putting together a report, how you create a flow for it and how it’s all connected. This is what makes a difference between the general LLM and what we do.”

His conclusion: “The general LLM will exist, and how each one of us taps into it and how we can manage the workflow to make it specific to our field, this is where the future is.”

6. Embrace it from the top, and onboard young talent with it

Greg asks what firm leaders get wrong about AI. Fakih points to skepticism and lack of know-how: “If you have a skepticism, it goes all the way from the top down, because you’re not going to let anyone use it. You’re going to have concern if anyone uses it. If you’re open to figuring out how to use it and train it and create guardrails for it, it will affect the whole organization as well. We see both, unfortunately more on the skepticism side. But I think this will change really quick, and it has to.”

He sees a gap between leaders and the people entering the workforce: “They have to embrace it, all the way from the top to the young engineer. The young engineers have been using it in college for their preparation and examination and figuring out how to study. They study through AI today. I know that for a fact. I’ve seen it. We mentor some engineers.”

At SEG, new planners learn the tools on day one: “We onboard some planners, and part of our onboarding process is to onboard them on our tools as well. One of them is ChatAEC early on. When they come in out of school, they spend so much time looking at websites and finding information. How much added value is this to them down the line in their career? I never heard anyone say, ‘I’m so glad I did all of this research for websites,’ because it changed every day. So to have this early on, it elevated their game. They’re getting the information, and now they’re working with the architect on the site plan based on the requirements, and they’re understanding the whole workflow.”

His message to his team and clients is simple: “It all comes from leadership. I encourage everyone dealing with me to use AI. I don’t care if they get an email from AI, as long as they’re prompting it correctly and they understand what it is and how to put it together.”

And he’ll tell you it’s already here: “We have to all embrace it. It’s a matter of time. My daughter is 10 years old. She did the business plan and flyers for school through ChatGPT. It took her 10 minutes to put together a business plan and create a marketing flyer. She printed 20 of them and took them to school. This is happening all around us, so we cannot deny it.”

Hear the full conversation

Watch or listen to the complete episode above, and subscribe to the AEC GrowthCast on YouTube, Spotify, or Apple Podcasts so you don’t miss the next one.

Want to see where technology is taking AEC? Join us at AEC INNOVATE.

SUBSCRIBE TO BLOG:
November 16, 2026

Hours Down, Value Up: SEG CEO Ali Fakih on Building AIAEC

This blog content comes from an interview conducted on October 13, 2026. For more information or context, watch the full episode where you get your podcasts. “We have a client..

Read More
November 16, 2026

Make Your Own Way:  Caldwell Architects CEO Miller Caldwell on Stepping Into Big Shoes

This blog content comes from an interview conducted on September 30, 2026. For more information or context, watch the full episode where you get your podcasts. “I had to make my..

Read More
November 10, 2026

How Better Employee Onboarding Supports Retention in AEC Organizations

Employee Onboarding for AEC Organizations: Why Is It Important for Your Retention Strategy? A new engineer can understand the technical requirements of an assignment and still..

Read More