by Cory Kamholz, AIA, LEED AP, NCARB

Left: Rhino massing model of a library project with no materials applied. Remaining four images: AI generated renderings that follow that same input geometry but also respond to a prompt describing the design and landscape.
AI in Architectural Design
Artificial Intelligence (AI) has transformed almost every profession in the world, offering surprising ways to enhance how we communicate, manage tasks and gather information. Through an array of AI-powered tools, the architecture, engineering and construction sector stands to benefit in myriad ways from emerging AI solutions.
With our sights set on continuing to offer our clients industry-leading design service, FGMA has formed an AI task force composed of designers, technical leaders and software and systems experts to test a range of workflows and available AI products and solutions. The task force is focused on four areas of investigation: AI as co-pilot, automation, enhanced visioning and graphics, and concept development.
Left: FGMA Original Project Rendering; Right: AI Enhanced Rendering
AI as Co-Pilot
Our project teams spend a lot of time getting to know our clients and the communities they serve. While there are no shortcuts to the research and solution-testing we do, we have found that AI can sort through large amounts of information quickly. Many of our clients have district or campus standards, and AI allows for natural language searches that are more effective than a string search and can quickly summarize longer documents. AI-assisted searches and summaries help us wade through lengthy documents, find information quickly and share it with the team. This allows our team to move forward with the confidence that we’re in compliance with our client’s standards.
Automation
In a recent tweet, fiction author Joanna Maciejewska (@AuthorJMac) wrote, “I want AI to do my laundry and dishes so I can focus on art and writing, not for AI to do my art and writing so I can focus on laundry and dishes.” Different versions of this sentiment from creatives can be found across social media platforms, and the world of architecture is no different. FGMA is testing solutions that make repetitive tasks quick and easy so we can focus on innovation and design in ways that make a positive impact for our clients. For example, tools that dimension a drawing or automate the setup of drawing sheets could give time back to the team that could be spent improving the quality and creativity of our work. Early tests show some promise, but there’s still no substitute for an experienced architect reviewing the work.
Enhanced Visioning and Graphics
Early in the design process, FGMA spends time working with each client to develop a vision for their project. This is a collaborative process, and when it’s done well, the vision represents a perfect blend of our client’s aspirations and our design team’s experience and design leadership. Sketches, massing models (both digital and physical), mood boards and precedent imagery are still useful, but AI provides another tool. Hand sketches, while sometimes beautiful in their own right, are not always easy to understand. Precedent images and mood boards can approximate a new idea, but they are not the idea itself. AI gives us the ability to turn sketches into realistic and accessible images and to generate mood boards and concept imagery that is more attuned to our design ideas and our client’s goals.
Left: FGMA Sketch/Veras Input; Right: AI Enhanced Rendering/Veras Output
Concept Development
As an extension of visioning, AI can be a useful tool in adding realism to concept imagery. A range of AI platforms allow us to use a 3D model or perspective drawing as part of the input prompt, creating the basis for the AI image in a way that allows the design concept to remain intact even while AI is given the freedom to fill in the details. Various controls in the AI give more weight to an input prompt, assigned materials or our substrate image, or we can give AI more freedom to introduce randomness into the process.
There is a risk-to-reward calculus that takes place when using AI as a concept development tool. The lack of precision in material selection and lighting means that these renderings are unlikely to represent the concept as imagined by the designer, and chasing precision can add more time to the process than a traditional rendering workflow. But there’s an ability inherent in AI to quicky generate and test many options and uncover related forms while we rely on the design team to evaluate which options are worth further investigation.
"The AI tools we’re testing need to be treated like inexperienced employees working under the design team’s direction." - Cory Kamholz, AIA, LEED AP, NCARB
Left: FGMA Original Project Renderings; Right: AI Enhanced Renderings
Sometimes the work the AI produces is interesting, surprising, unique, useful or all the above. Other times, the results are incomplete, inconsistent with the prompt or just plain wrong. As these tools continue to mature, their accuracy will improve and the list of tasks they can be trusted to accurately complete with limited oversight will expand. FGMA sees the potential in developing new workflows that allow us to offer more to our clients, especially as part of the visioning and early design process. A helpful database of AI tools (along with some training resources) that support the AEC industry can be found at www.aiinaec.com.
Cory is a design studio leader at FGMA with more than 22 years of experience including programming, master planning, design and design leadership across a variety of markets, with a focus on higher education. He brings a passion for sustainability and design excellence to every one of his projects.
This article emerged after AI and FGMA took center stage in a Construction Owners Association of America (COAA) workshop in October 2024 at the University of Wisconsin-Milwaukee. The afternoon included a series of speakers followed by a panel discussion, all focused on the impact AI is making in the design and construction industry.