2026 AEC trends (Part 2): AI in architecture—vibe coding, real-time analysis, and data
February 19, 2026
In a year where AI is supposed to change everything, how do we keep up?
What will 2026 bring to the architecture, engineering, and environmental design industries? Change is all around us—and in architecture, we are squarely in the “hype cycle” of artificial intelligence (AI). But it’s also immensely exciting.
This is part two of a three-part series examining the trends that will shape the architecture, engineering, and construction (AEC) sector this year. Part one looked at nature-based solutions in environmental work and power delivery for energy teams.
For architects, it’s a tricky balance to use AI effectively without thinking it’s the answer to everything, as I’ve discussed before. It is one of many modern tools (such as computational design and real-time visualization) that allow us to deliver beautiful designs in a twenty-first-century process. However, at the pace it’s evolving, AI will likely have a bigger, faster impact than any of these tools.
For me, here are three trends that will define AI in architecture in 2026:
- “Vibe coding” custom tools for automation and rapid prototyping
- Real-time analysis and simulation-driven design tools
- Standardization and ownership of data
As part of IdeaXchange, an internal design competition, teams used AI tools to create a North American headquarters of a fictional aerospace company. (Pictured: Exterior view of Team 7’s winning entry.)
1. Vibe coding is the new digital sketch
For me, this is the single most interesting trend for 2026.
First off, what is vibe coding? It’s the practice of asking AI to build (i.e., code) a new piece of small software. It uses natural language prompts, which makes it great for designers—who can explain what they are trying to accomplish a lot faster than deciding what coding language to write it in. We’re not writing code line by line; we’re simply guiding the AI agent to create this piece of software.
The AI technology can produce ready-made software solutions to solve very specific design challenges. It allows designers to create custom tools in hours—or even minutes.
Most AEC companies cannot support dedicated software development teams to create new AI tools. With vibe coding, we don’t need to.
With only a basic understanding of coding—something many design technologists already have—we can create targeted, functional tools on demand.
Here’s an example.
Imagine I’m designing a building façade and want to study different louver layouts to block sunlight at different times of the year. How long are the louvers? How tightly are they spaced? How are they arranged?
With vibe coding, a designer can write a piece of software that models the louvers and then analyzes them. In one afternoon, the team now has custom software to solve this challenge. They can conduct analysis and rapid prototyping before moving into a more detailed study of a selected option. Think of it like a digital sketch, where AI is the pen.
I believe this gives you a glimpse into the future of computing, where commands are made through natural language, not by button clicks.
Having AI handle some of the work for us helps, but it hasn’t greatly sped up the overall project yet. We’re talking about a small portion of a huge building design project.
Is this new tool designing the whole building? No. Did it draw the complex details for you? Again, no. But it’s letting us create a custom tool to rapidly ideate based on technical requirements—all with a very low barrier to entry.
This is available to everybody via Copilot, ChatGPT, Claude, etc.
For 2026, vibe coding will be the big thing for AI and generative design. But in two or three years? Hard to say.
Let’s look at it this way: Three years ago, few believed AI-generated films were possible. Today, there is an AI actress—Tilly Norwood, created by a UK AI production studio. In the music industry, AI artists are topping the charts. Things are moving fast—and the skeptics are regularly proven wrong.
Right now, these tools are bespoke and challenging to scale. But maybe that isn’t the point—this is about every designer having the power of a small software team at their fingertips, ready to solve our everyday problems.
Does it speed up the design/construction process? I don’t think so; at least not yet. Let’s go back to the louvers for some insight.
Having AI handle some of the work for us helps, but it hasn’t greatly sped up the overall project yet. We’re talking about a small portion of a huge building design project. But you’re more confident in the performance of your designs, you’re able to rapidly prototype multiple options in a short time, and you’re more practiced in the art of controlling this new AI technology.
2. AI allows for real-time analysis and simulation-driven design tools
Getting real-time feedback in an industry where analysis software is traditionally slow opens questions about the future of project timelines.
Here’s an example. My colleagues Mo Elsayed and Justin Shultz are creating real-time analysis tools that are operated using natural language. They’ve written about it previously:
- Generative AI, designing at the speed of innovation
- Rethinking energy modeling: From bottlenecks to breakthroughs
- The decarbonization imperative: A shared mission in AEC
As a designer, you might say to the tool: “What is the optimal orientation of my rectangular office building on my site that maximizes daylight and minimizes energy use?” The tool will do basic energy modeling based on the information you provide it.
Vibe coding is one of the top trends for AI in architecture for 2026. (Pictured: Team 6’s entry for IdeaXchange.)
A lot of the analysis we used to do around building energy performance had slow turnaround times. The architects would hand over a model, and several weeks later, you would receive feedback on whether the building could perform as needed.
Now it’s two weeks. Or a week. Sometimes it’s on the same day. And soon, with the help of AI, our designers can get building performance feedback alongside their design within minutes. That is a monumental shift.
The result?
We can design more energy-efficient buildings, because the work becomes easier and faster. We can simply talk to AI with natural language—and get that real-time analysis to help move the project forward.
With the reduction in time—and associated costs—you can now have a higher-performing building at the same cost. Maybe in the past, the client didn’t have a budget for energy modeling. Now it’s on the table. So, as an industry, we’re able to produce more energy-efficient buildings.
And this takes us back to that louver example from earlier.
If you go back in time, people would have been drawing these sections with a pencil, getting the solar azimuth, checking the lines and angles of the sun at different times of the day.
Now, we can do that whole process in 3D and almost instantly. And that gives us a higher-performing, more evidence-based building.
3. Data life cycle: Harvesting, standardization, and ownership
Here are the skeletons in the closets: Many firms are now facing the reality that their data is messy and can’t be used for deep analysis or for training AI.
This is likely the single biggest barrier to implementing AI and generative design.
Many firms are finding that they have unorganized collections of drawings. They likely have Building Information Modeling (BIM) models and project management information that are not formatted to a common data structure. The accounting software is separate from their resourcing software, which is separate from the design software, etc.
AI allows for real-time analysis and simulation-driven design tools for architects and designers. (Pictured: Interior view of Team 7’s entry for IdeaXchange.)
While AI can help clean it up and give the data life cycle a kick-start, we all just need to organize our data better. We must ask: How do we organize our data so we can let AI do its best work? What are the fundamental steps required to create data and a culture that can eventually be empowered by that data?
For our teams in the Buildings practice, we are tackling it from two directions. First, we are harvesting every piece of information in the firm, whether it’s a structure or not, and putting it into searchable databases. Second, we are developing foundational tools to set standards for our digital assets. And that makes the data useful in the future. Sometimes, this work may not sound revolutionary—like creating strict BIM modeling protocols for all new projects—but without a new heightened level of care for data, a firm is likely to fall behind.
This leads me to another big challenge in the age of AI—ownership over intellectual property. Traditionally, designers own the trademarks for their designs; many modern contracts transfer that ownership to clients. Additionally, we are seeing terms of service from software companies that claim ownership of the data flowing through them, effectively asserting ownership over designers’ drawings.
These kinds of structures could be catastrophic for designers, limiting their ability to train AI models based on their designs at a time when automation may drive down design fees. The conversation around ownership is sure to hit fever-pitch over the next year.
For 2026, it’s about adjusting to practice and model shifts
We are seeing a rapid expansion of digital tools in an industry traditionally slow to process change. Keeping up is about more than simply buying or developing software; it’s about larger shifts in how we work. Our industry will need to open up to new career paths for technologists, new ways to assign value to career performance (time saved vs contracts won), and strict data care. All the while, we need to think deeply about what the future of our industry would look like if projects took half the time to deliver.
It’s an exciting time to be a design professional, and I look forward to making sure our design professionals remain ahead of this rapid technological change.
