.svg){kind=icon}
On January 11, 2024, I joined the Urban Land Institute (ULI) Tampa Bay – January StimULI event to chat about AI in real estate development on behalf of TestFit. It was a fantastic panel discussion with a few innovators in prop-tech:
- Wade Vaughn -Senior Manager, Strategic Property Partners
- Tyler Carlson - Co-Founder, ReSquared
- Sabrina Dugan - VP Research and Development, AutoReview.ai
- Spencer Muratides - Co-Founder, EarthNotes.ai
- ULI Tampa, Leadership of Lee Lowry
Here are my 4 takeaways from the conversation:
AI as the Ultimate Intern
Wade posed the question, “Where do you see AI impacting commercial real estate, and how might your company incorporate AI into its business?” The answers were fascinating and all centered around the same concept—the Ultimate Intern.
Tyler coined the term, Ultimate Intern, and referenced Gong which removes the need for you to take meeting notes and assists users with follow-up tasks. This type of artificial intelligence (AI) doesn’t replace anyone but makes users much more efficient by freeing up their time.
Spencer brought us all to reality with the real estate tech stack. Most of us use some combination of Excel, Outlook, and some sort of site prospecting software. He argued that there’s a lot of room for improvement in our current real estate workflows. It’s exciting because development firms can become more efficient and remove some of the busy work that takes away from us getting more buildings built.
I agree with both of them and argue that AI in real estate, in its current path is going to push us into the future. The ones that do adopt new AI tech early will be the ones with the best deals, best returns, best projects, and best employee retention. The Ultimate Intern means employees are focused on what they do best, not counting parking stalls or sifting through folder trees.
Real Estate Data for AI is Tricky
Remember the days when startups all had a free product with no monetization strategy but they had oodles of data so they could “monetize” that? It was a push to collect, collect, collect. Well, now we are transitioning to using that data effectively. AI is really the only way to do that. But it has an existing problem, old data isn’t clean.
We had a question from the audience about this, “When the data that the model is trained off of is old and untrustworthy, how do we generate new data to improve the model?”
We’ve already seen some AI generators come under fire because they aren’t connected to live data and are only referencing old data, whether that’s outdated, inaccurate, unclean, etc. It creates problems.
Another question from the group was about how that applied to real estate developments, “How do we get more reliable comps?”
Spencer noted that everyone wants as much data as they can get but wants to share as little as possible. He also quipped it’s why we see errors in some of the comps data providers have because no one wants to share their real data.
Getting to an end state of clean, accurate data for using AI in real estate development will be difficult because we don’t want to share with each other. However, as Sabrina noted, services that provide niche-focused large language models (LLMs) could help solve the issue. There may be a future where data is shared or a well-trained model that generates hyper-accurate data to help train itself. Lots have to happen before we reach that future but it’s possible!
The Future is Bright for Analysis
Sabrina’s leadership here with autoreview.ai was fantastic. She shared that AutoReview works with machine learning to import construction documents and verify their accuracy with local zoning codes. Their AI can count parking stalls, check setbacks, look for errors in documentation, and more. That type of analysis can take weeks for cities.
But with the help of AI, the planning department can now accept files and have them fully reviewed in hours. What an unlock for cities and developers!
Big analysis takes time and expertise. Soon the days of pivot tables and sql will be gone and AI will do all the analysis for you. Imagine importing your data from all your past as-built projects with construction type, units/layouts, rents, and lease-up time, and then AI analyzes all of it for you. TestFit could then recommend the optimal layout, kit of parts, and rents for your real estate feasibility study. What a bright future right?
Elevating the Artist in Real Estate
Wade got us all thinking, what happens to the artist with the addition of AI? It made me think, what if you trained a model on all of Picasso’s paintings and then started creating new paintings from it? Does that diminish his artwork or does the new work hold a similar value as the past because it’s “from the mind of the artist”?
We had lots of opinions. Spencer talked about how original artwork increases in value and makes great artists even more sought after.
I also talked about the artist in real estate development—the architect. Our founder Clifton Harness was the artist in a development firm and struggled to see how counting parking stalls qualified as art. Just like how AI will be a great intern, it will be a great assistant to the artist. With TestFit, Clifton envisioned a world where he could automate the repeatable and tedious tasks so he could focus more on the design, the real art.
This world is fast approaching. TestFit already helps thousands of real estate artists speed through the mundane and build better deals faster.
But what about the future? Could we get to a point where AI enables us to build more beautiful buildings within budget? What if AI frees up the architect to design buildings that emulate our ancestors' homes from Europe but keep the project moving fast and on budget using 3D printing and AI construction on-site tools? We could be on the cusp of a more beautiful future.
We'd love to thank Lee for her leadership at ULI Tampa in putting together this event around AI in real estate. If any other chapter of ULI would like to host similar events, we will be there!
