Three years ago, executives at Gensler launched an internal experiment they called the AI sandbox. The goal was simple. Test generative artificial intelligence tools with open curiosity. Avoid knee-jerk reactions. Learn what actually worked inside one of the world’s largest architecture practices.
“It was a key moment for us as a firm,” Jordan Goldstein, co-CEO of Gensler, told Business Insider. “We knew we couldn’t wait and be reactive to artificial intelligence — we had to be ahead of the curve.”
The timing proved sharp. Gensler handles roughly 3,000 projects annually. Today AI touches the majority in some form. Designers simulate sunlight patterns in hallways. They model sound propagation across open offices. They map human traffic flows long before ground is broken. These capabilities compress what once took weeks into hours.
Early tests involved vendor tools. Some graduated into custom interfaces built in-house. One standout example arrived six months ago. RunDiffusion, a platform for generating images and videos from prompts. Within four days more than 3,000 Gensler staff completed training. Co-CEO Goldstein himself produced a viable concept in minutes after basic prompt engineering instruction.
The payoff shows in client presentations. Before bidding on a mixed-use district tied to a new stadium, the firm turned static designs into fluid cinematic sequences. “The biggest breakthrough has really been in storytelling,” Goldstein said. “We realized we could take AI-generated design concepts and turn them into very compelling cinematic narratives.”
Similar techniques shaped the vision for Baghdad Sustainable Forests. The project, billed as Iraq’s largest ecological development, demanded clear depictions of future landscapes. AI helped translate abstract goals into vivid, testable scenes.
At Under Armour’s forthcoming Baltimore headquarters, the firm went further. AI modeled daily employee movement patterns. It predicted occupancy impacts on ventilation and energy loads. Such simulations let teams iterate on environmental performance without physical prototypes.
Gensler’s 2026 Design Forecast frames these experiments inside larger industry currents.
Released in December 2025, the report identifies six forces remaking real estate and cities. One stands out. AI functions as more than a speed enhancer. It acts as a creative partner that exposes hidden patterns in how humans experience buildings. The forecast argues this shifts competition from pure efficiency toward sustained innovation. (Gensler).
Co-CEOs Goldstein and Elizabeth Brink captured the convergence. “Design has always been about imagining what’s next. Today we’re seeing an extraordinary convergence of technology and creativity that’s expanding what’s possible. AI isn’t just accelerating our work — it’s revealing patterns in human behavior we’ve never seen before.”
Experience now trumps square footage as the primary measure of property value. Immersive districts anchored by sports venues succeed because they deliver emotion and narrative. Agility in design becomes essential amid volatile costs and supply chains. Existing building types find new lives. Malls convert to universities. Transit hubs host exhibitions. And climate resilience moves from optional to non-negotiable. (ARCHITECT Magazine).
Gensler’s Global Workplace Survey 2026 adds texture. AI power users — employees who deploy the technology regularly in both personal and professional settings — behave differently. They spend less time working solo. They devote more hours to learning, socializing and virtual collaboration. These patterns suggest AI does not isolate. It can redirect energy toward distinctly human activities.
The survey, covering more than 16,000 office workers across 16 countries, found stable overall habits. People still allocate about 40 percent of their week to solo work, 27 percent to in-person collaboration. Yet the subset most fluent in AI reports stronger team relationships and greater engagement in skill development. Over 80 percent of learning-oriented employees say the physical office boosts productivity, satisfaction and collegial ties. (Gensler).
But adoption carries friction. A 2025 American Institute of Architects study showed roughly 90 percent of architects harbor worries about inaccuracies, data security and blurred lines between machine and human output. Professors echo the tension.
Jason Vigneri-Beane at Pratt Institute sees value in early conceptual phases. “I find that for everyone who is excited about it, there are also people who are concerned about losing agency and authorship, as well as energy and water use,” he told Business Insider. He reserves AI mainly for experimental aesthetics and testing conceptual possibilities.
Sabri Gokmen, assistant professor at the University of North Carolina at Charlotte, integrates the technology into data analysis for cultural heritage projects. He also teaches students image generation and prompt techniques. His current collaboration with computer scientists aims higher. They train custom models that accept text or image prompts and return actual 3D geometry — something commercial large language models still struggle to deliver reliably.
“We are trying to make it in a way that someone like an architect or designer can write prompts or give images, and what they’ll get back is a 3D model,” Gokmen said. He remains optimistic. “In terms of design, architecture will still rely on human decisions and social contracts. But this technology will open up other avenues for growth.”
Recent industry lists reinforce the momentum. Publications catalog dozens of specialized tools for 2026. Midjourney and Adobe Firefly dominate early ideation. Veras and ArkoAI handle rendering inside modeling environments. Generative planning platforms such as Maket, ARCHITEChTURES and TestFit accelerate feasibility studies. Energy modeling and code-checking features continue to mature. (Chaos, published five days ago; Snaptrude, updated June 22, 2026).
Gensler’s internal course offerings and partnerships mirror this breadth. Design Director Aubin Gastineau leads a six-week program for architects that covers workflow integration from concept to optimization. Students work with Midjourney, DALL·E, Firefly, Finch3D, TestFit and ChatGPT among others. The emphasis rests on responsible application and systems thinking. (ELVTR).
So what does this mean for practice? AI compresses iteration cycles. It multiplies options without proportional headcount. It equips teams to present living, testable stories rather than static boards. Yet the human role sharpens. Judgment on aesthetics, ethics, client needs and long-term societal impact stays firmly in designer hands.
Goldstein’s early sandbox bet looks validated. Gensler did not wait for the technology to arrive fully formed. The firm invested time in experimentation, training and integration. The result appears in faster concept validation, richer client dialogues and designs that respond more precisely to environmental and behavioral data.
Other firms watch closely. Some experiment cautiously. A few still hesitate. The gap between early movers and laggards may widen as simulation accuracy improves and 3D generation matures. For now, Gensler’s blend of sandbox pragmatism, forecast vision and workplace research offers a practical map. Test relentlessly. Measure human outcomes. Keep creativity at center.
And the buildings that emerge? They stand to feel more attuned to how people actually move, learn and connect. That alignment, more than any single algorithm, will determine which projects endure.
WebProNews is an iEntry Publication