Look at a vacant lot in a growing or desirable location, and you may be looking at a problem. The easiest parcels have already been developed. What remains are sites with unusual topography, drainage challenges, or buried infrastructure that complicates excavating earth and building foundations. Turning these parcels into successful developments requires skills and experience.
“There’s a reason it’s not been developed at this point in time,” says Josh Thompson, digital practice specialist at Stantec, one of North America’s largest engineering firms. “Because there is some sort of constraint that is making that lot hard to develop.”
Thompson is talking about site development, but he could also be addressing his own industry. In civil engineering, the most significant constraint isn’t land, it’s people. The field is facing a “silver tsunami” of early retirements—just as engineering firms are struggling to find new talent. “The world needs civil engineers now more than ever,” according to Feniosky A. Peña-Mora, president of the American Society of Civil Engineers. He estimates that over the next decade, “approximately 17 million infrastructure workers will need to be replaced, which is more than the current entire infrastructure workforce.” That’s one reason why in 2024, Stantec became an early adopter of Bentley Systems’ OpenSite+, a design solution powered by artificial intelligence (AI) that helps to automate time-consuming and redundant portions of a site design.
Stantec has a lot of work. But to start that work, it needs engineers skilled in one of infrastructure’s foundational disciplines: site grading. The name describes the intricate process of sculpting terrain to accommodate buildings, parking, drainage, and utilities while minimizing the expensive work of moving dirt. “We need more people than we can find for the skills that we require,” Thompson says.
Thompson is supporting roughly 2,000 designers and engineers using CAD software in Stantec’s 34,000-employee organization. From his perch, he clearly sees the bottleneck and also a solution. “Teaching somebody how to grade a site is extremely difficult. You kind of either get it or you don’t,” he says. Even experienced designers might take years to develop the intuition for optimal cut-and-fill, balancing aesthetic appeal with construction cost and regulatory compliance. With OpenSite+, Thompson says, “we’re going to be able to get more folks able to grade.” The software can generate site design proposals, but Stantec’s engineers remain firmly in control, reviewing, validating, and refining the results before they move forward. The shift allows experienced staff to focus on engineering judgment rather than manually producing every design iteration.
Testing the Future of Site Grading
The promise of what’s possible began to take shape last July during a multi-day working session in Sarasota, Florida. Stantec engineers spent time with Bentley’s OpenSite+ product management and development teams exploring early capabilities, testing the software, and discussing how AI-enabled site design could evolve over time. Using real project context, engineers worked through demonstrations, tested features, and shared feedback on desired functionality, usability considerations, and broader industry pain points.
The software uses AI to change how civil engineers design sites for construction projects. The platform brings grading, earthwork analysis, drainage design, regulatory requirements, and other tasks into one workspace, letting teams quickly test site layouts and weigh costs against environmental impact. AI performs much of the underlying computation and analysis, allowing engineers to focus on reviewing and evaluating results rather than manually producing every component.
“I’m passionate about this work because I’ve been applying artificial intelligence to civil engineering for almost 20 years,” says Ron Breukelaar, a distinguished engineer at Bentley. “That’s why OpenSite+ is so close to my heart. I believe we can greatly improve the civil engineering and site design world by creating a product that is smarter, eliminates much of the tedious work, and improves the overall workflow for engineers.”
During the Sarasota sessions, engineers explored AI-driven grading optimization capabilities, including how the software generates and evaluates multiple grading scenarios based on different priorities. The discussions focused on how these optimization approaches could support engineering judgment by helping teams assess tradeoffs earlier and more consistently within the grading process.
OpenSite+ is also being designed to account for real-world constraints such as local regulations and regional construction economics. “The price of moving dirt in one area of the country is not the same as in another,” Thompson explains. In some markets, contractors install retaining walls to minimize grading. In others, walls are cost-prohibitive, so designers spread cuts and fills across larger areas. OpenSite+ allows users to adjust unit costs so designs can be evaluated within those regional realities.
The same principle applies to utilities. Water lines must be buried at least 3 feet deep in Texas to prevent freeze damage. In Calgary, Canada, where Stantec has a significant presence, frost depth requires 8 feet of cover. These aren’t academic details—they’re constraints that fundamentally affect site design, and AI that ignores them produces unusable results.
The combined insights help users find “the best model for that region,” Thompson says.
Building in Three Dimensions
There are other benefits. Thompson says that traditional CAD software has designers “just drawing lines and circles” in two dimensions, then relying on separate calculation tools to determine elevations and volumes. OpenSite+ builds a three-dimensional model of the site from the start, utilizing real‑world, smart objects that can report quantities for both the sculpted terrain and the infrastructure that will be constructed. That way, engineers don’t have to rely on unintelligent CAD lines used to display contours and the outlines of construction elements.
For younger engineers without decades of 2D CAD experience, this may prove liberating. “I do believe it’s going to be relatively intuitive and easier to get some of our newer staff to onboard,” Thompson says.
All of this doesn’t mean that AI has the final word. Experienced engineers review AI-generated designs before they get passed on to the next team. This is a different cognitive task than creating site designs from scratch. Using AI to come up with the most viable designs allows engineers to be both faster and more thorough. Site grading errors are expensive and catching them in the design phase rather than during construction can save tens of thousands of dollars on a single project. “For years, engineers have been producing deliverables to support projects, drawings, documents, and reports,” says Ian Rosam, Bentley product management director. “But if we can automate and streamline these deliverables down to the push of a button from the model, wouldn’t that be fantastic? That to me is the mindset change for digital delivery. It’s linking and automating the traditional deliverables of sheets with the models in a unified way, so they can work seamlessly together and provide greater project insights.”
The Feedback Loop
As OpenSite+ early adopters, Thompson and his team have participated in bi-weekly meetings with product managers and developers for nearly a year, shaping the solution’s evolution.
That collaboration was apparent during the Sarasota session. “We were making comments and feedback, and some of the items were put into an update just for us to test that next day while we were there,” Thompson recalls. Issues identified that Monday morning appeared as fixes Tuesday. “Everything we’re doing is new,” says Bentley distinguished engineer Scott Devoe. “The AI stuff that we’re doing is new. I’m looking in the mirror because I still feel like I’m 20.”
This iterative development reflects a strategic choice by Bentley: Focus first on industrial and commercial site development, including complex facilities such as data centers, while building on a strong AI foundation. Municipal projects, such as schools, are also part of that focus, with residential developments presenting a different set of challenges to be addressed over time.
Stantec is planning to integrate AI-generated designs into its standard workflows, quality control processes, and the designs it delivers to clients. The transition has been a catalyst for growth. After more than 20 years working with legacy software, Thompson describes the shift as an opportunity to rethink ingrained workflows and embrace a more modern approach. He’s energized by OpenSite+, describing it as “the most advanced software in the site design ecosystem.” Thompson adds, “Y’all have created a tool that is more advanced than anybody else’s. They’re going to need to catch up to your software.”
Thompson still keeps his drafting scales at his desk—relics from the days when engineers drew blueprints on paper. They’re a reminder of how radically the tools have changed while the fundamental problems remain constant: making sites work, on budget, on schedule, with the people available.
FAQ:
Site grading is the intricate, foundational process of sculpting a parcel of land to accommodate buildings, utilities, and parking while minimizing the costly movement of dirt. It is notoriously difficult to teach and can take years for designers to master, making it incredibly hard for engineering firms to find enough talent with the right intuition for the job.
OpenSite+ leverages artificial intelligence to automate the most tedious, time-consuming aspects of site design, bringing grading, drainage, and earthwork analysis into a single 3D workspace. Instead of manually drawing 2D CAD lines from scratch, engineers can use the software to generate and evaluate multiple grading scenarios, freeing them up to focus on high-level review and engineering judgment.
Yes, OpenSite+ is specifically built to evaluate designs based on regional construction economics and local constraints. For example, it understands that a water line only needs to be buried three feet deep in Texas to avoid freezing, but requires a massive eight feet of cover in Calgary, ensuring the AI generates usable, region-specific models.
