This episode features IMEG civil project manager Matt Pohlhaus in a discussion on how artificial intelligence is transforming site design. Based in the Washington, D.C., metro area, Matt leads land use and civil engineering projects across Maryland and West Virginia. Increasingly, he says, AI is becoming as much a part of his toolkit as CAD software or site surveys. “We use artificial intelligence daily,” Matt explains, describing how it’s woven into tasks both big and small—from communication to design. “If you’ve ever been stuck trying to get some language out the right way, just throwing a few prompts into ChatGPT or something similar” can result in a “very well-worded email” and freeing up time, he says.   On the conceptual side, his team is utilizing AI-driven generative design software. With just a site location and a few inputs, the program quickly produces fully fleshed-out site layouts. “When a client asks, ‘Can we put a 60,000-square-foot grocery store on this site?’ I can now show them in minutes,” Matt says. In the past, that answer might have taken days of drafting and another round of meetings. The ability to test ideas in real time with clients has proven invaluable. “The coolest thing about it is everything updates on the fly,” Matt says. He describes meetings where clients ask to move a building across the site or add a parking garage—what once required rescheduling is now an instant adjustment. “It becomes a lot more conversational,” he says. “I think clients tend to see us more as a partner than just a consultant drawing lines on a screen.” A medical office building project, for example, completely shifted direction during a single meeting. The client had arrived with a looping driveway design they thought was final, but after moving the building within the AI model, the layout quickly evolved into something more straightforward, visible, and cost-efficient. “That was probably a 20-minute conversation,” Matt says. “And the scheme they ended up moving forward with was completely different from what they came in thinking they were going to do.” Another project—an industrial site tied to a rail line—showed off the software’s deeper analytical power. The developer wanted to run railroad tracks into the property, but when Matt layered in topographical data, a problem appeared immediately: the proposed line ran over a 30-foot cliff. “If anyone’s been on a train before, they don’t go up and down hills all that well,” he says. By shifting the entry point half a mile, the team avoided an impossible design and a change that in the past might have taken weeks of back-and-forth.