Newsletter
Why Every CRE Firm Can Be AI-Native Starting Right Now
Why CRE's light legacy stack is now an advantage and the practical playbook firms can use to become AI-native today.
Commercial real estate firms are historically the slowest to adopt tech. That's about to become an advantage.
CRE teams run on a bare bones tech stack--Microsoft Office suite, CoStar, and an extra data provider or two. That's typically it.
For a long time that looked like a liability from the outside. Right now it looks like an advantage.
Firms that built on legacy systems ten years ago are now scrambling to AI-proof everything. Firms that never built on anything have a clean slate.
Most firms I talk to have that clean slate. You don't have to migrate anything. You don't have to convince a big committee. You're just deciding: what's the best way to do this, today, with the tools that exist?
What AI-native actually means
It's not about converting your deal data to a SQL database. It's not about hiring an ex-FAANG engineer to run your AI task force. It's about culture and willingness to change your process.
AI-native means starting again from first principles: What work needs to get done? What is the most efficient way to do it? And, do we actually need this output--or did we just inherit it?
That last question changes everything.
Instead of asking AI to populate your existing underwriting model, you ask whether that template should exist in its current form at all. Instead of hiring a pure CRE analyst, you hire someone who's 70% AI-fluent and 30% real estate guru--because that ratio is what the next generation of this work actually demands.
It's not about automating full processes end-to-end with no human review - it's about providing compounding 20% advantages that allow you to out-scale competitors.
Here's the playbook I've seen work
- Tear apart one process at a time. Pick something your team does every week--deal screening, IC memo prep, comp pulling. Don't just say "we need to improve underwriting." Break it into five steps and find where the inefficiencies actually live. See an example of this in the Appendix.
- Hold a weekly 60-minute AI standup. Everyone shares what they used AI for that week--what worked, what didn't. Learning in public with your own team shifts culture faster than any training or rollout.
- Run an internal competition. Who can build the best workflow automation? Who finds the best tool? Put money on it. The goal isn't just the output--it's building the muscle of looking for these opportunities in the first place.
- Pilot multiple tools at once, knowing some won't stick. 90%+ of the deals you underwrite don't pan out. Apply the same logic here. The goal isn't a perfect stack. It's figuring out what tools work the best for your process.
The firms that move in the next 12 months are going to build a process and culture advantage that compounds.
We talk to CRE investment firms every day who are figuring out what AI-native actually looks like in practice. If you want to think through what it means for your team, book a call here.
Appendix
A Real Workflow Audit - Underwriting
Every acquisition team underwrites deals roughly the same way they did five years ago. Below is what it looks like when you actually pull the process apart. We've run this with enough teams to know the findings are almost always the same.
Step 1: Deal intake
A deal comes in and someone decides whether it enters the pipeline and how it gets logged.
What we typically find: A senior person reads the email, makes a gut call, and either forwards it to an analyst or lets it sit. If it gets logged, the analyst re-types information that already exists in the original email. There's also no standard for what "enters the pipeline" means--one analyst logs everything, another only logs deals that clear an informal mental screen.
Question to ask: Are you saving down and tracking every deal on the market?
The takeaway: Write down your intake criteria first. Then automate the logging. A deal that comes in via email shouldn't require a human to re-type it into a pipeline.
Step 2: Quick screen
The deal is in your pipeline. Now you need to decide if it's worth pursuing.
What we typically find: Analysts spend one to two weeks gathering basic information before the first meeting to discuss it. By then, someone else has already moved.
Question to ask: How long does it take from a deal entering your pipeline to having a clear picture of whether it's worth pursuing?
The takeaway: Once your intake criteria are set, this step should be automated. A quick screen should surface the basics on any deal within hours, not weeks--so your team walks into the pipeline meeting already knowing whether it's worth the next step.
Step 3: Spreading financials
An analyst pulls the financial package and populates the underwriting model.
What we typically find: Entirely manual, highly variable. Financial packages arrive in every format. A clean package takes two to three hours to spread. A messy one takes a full day.
Question to ask: How important is it that you follow every normalization step, can you let the AI run freely for an initial underwrite? Would you underwrite more deals if you could automate this step?
The takeaway: Document your normalization rules before you automate anything. There are both specialized and general tools that can help here--and this is one step where letting AI run with a little less precision is often worth the time saved.
Step 4: Market research
The analyst pulls comps and builds the market context behind the underwriting assumptions.
What we typically find: Highest variance of any step. Some analysts go deep; others pull three comps and move on. Research is almost never reused--an analyst builds a Phoenix comp set, six months later someone else builds the same one from scratch.
Question to ask: Are we hitting the same data sources every time? How often are we actually using internal data we've already pulled?
The takeaway: If your analysts are running the same CoStar searches on every deal, that's something you can automate. And research that gets produced should be stored somewhere your team can find and reuse it--not buried in a deal folder nobody opens again.
Step 5: Presentation
The deal gets packaged for IC, a partner review, or an investment memo.
What we typically find: Format varies by deal, by analyst, and by who it's for. Mostly manual, minimal templating.
Question to ask: How long does a cookie-cutter deal take to package? Are the customizations actually necessary?
The takeaway: Not every deal needs a perfect memo. Standardize the format, cut the layout decisions, and build a process that takes outputs straight from the model into a presentation. The analyst's time should go toward the analysis, not the formatting.