
Welcome to the CRE AI Studio newsletter. Your weekly dose of AI news that matters and tips on putting it to work in commercial real estate.
Quick Heads Up: You can join us on a 7-day free trial to CRE AI Studio here.
AI Platform Updates You Should Know About
OpenAI's GPT-6 Astra started rolling out September 3 to a limited set of organizations first. Plus, Pro, Business, and Enterprise get access over the following days. OpenAI says it can run longer multi-step jobs without you steering every click. It builds polished documents, spreadsheets, and presentations, and is stronger at fixing real software. For CRE: describe a house in plain English. Astra can drive 3D tools already on your computer. OpenAI's demo used Blender for an editable model and camera tour, then Unreal Engine (interactive walkthrough software) so you can walk the rooms. Treat that as supervised concept viz, not construction drawings.
ChatGPT Sites lets you build and host a dashboard or small web app from a prompt, without standing up your own website. OpenAI is rolling out email invites so you can give named people viewer access without making the Site public. They sign in with the invited account. That means a private lead-tracking or deal dashboard you can hand a client. Catch, per OpenAI: Sites cannot connect live to outside data sources yet. To refresh numbers, pull from connected apps in chat or a scheduled ChatGPT task, then update the Site.
Where AI is Moving the Needle in CRE
EliseAI is the housing and facilities proptech valued above $1 billion. It launched Apollo on September 3. Elise says Apollo is a single agent that can do anything already on the Elise platform, in text or voice. A leasing agent can say "message everyone who toured yesterday," and Apollo sends it. It can also pull portfolio data, draft renewal outreach, or open a maintenance work order, using only that person's existing permissions. First Montgomery Group's owner is already using it, per Elise.
A retail vacancy pitch used to lead with square footage, asking rent, and demographics. More owners and brokers now stack visit data, card sales, point-of-sale feeds, tenant credit, and site forecasts. The argument becomes: this retailer will sell here. Newmark Merrill's Sandy Sigal used AI that way to rebuild a pitch a retailer had already rejected. Humans checked it, then the retailer closed. Brixmor is wiring the same idea into its own leasing data. Catch: the stacks estimate. They do not replace judgment.
Re-Leased, a property-management software company, surveyed commercial landlords in the UK, US, New Zealand, and Australia. It says active AI use nearly doubled year on year, with the US and New Zealand both above 60%. The same whitepaper cites JLL: 90% of CRE firms are piloting AI, but only 5% say they achieved all their program goals. Landlords in the Re-Leased survey named lease administration and abstraction as the top use (60%). Re-Leased sells AI for that work, so read its own figures as vendor research; the JLL cite is the independent one worth remembering.
Get the AI news that matters for CRE + tips on how to use it, every week.
Use Case: an 80-unit DD checklist, done in an hour
Sunday night I was going through email and a due diligence list came in on an 80-unit building we're selling in Manhattan. Everyone in this community has seen one of these, whether you're buying, selling, or brokering. I decided to run the whole thing through Claude Code instead of parceling it out across the team.
Two steps.

First, I opened the DD spreadsheet and added a column, highlighted in pink, with commentary on where each item lived. Four buckets. Dropbox, where we keep most of our files. AppFolio, our PMS, which I have the API keys for. Things Claude had to build from scratch, like a 180-day lease trade-out report. And things I had to ask Property Management for, which I had Claude draft and send the emails for.

Second, the prompt. Here is a version of it:
We are in talks to sell property x. A due diligence list was sent over, which can be found in the ad hoc subfolder, in the DD subfolder (there is a DD spreadsheet in there). Please confirm that you see this. I'd like you to work through that spreadsheet and create subfolders for each line item in there. Then let’s create a plan that has you pull all of the information being requested and puts them in those right folders. Please ask as many clarifying questions as possible. I’m looking at the list now- it's either going to be in AppFolio, Dropbox or asking APEX or me to pull it. I added commentary in column c that you should follow. Let's plan this out. The goal is to have all of this information in the subfolders this week.
Twenty-five minutes later, every line item had its own subfolder, filled. Reports it built. Marketing photos. Title reports. Leases. Financials pulled directly out of the PMS.


The human review is still on me and my team, and that's the piece that will take three days. But the aggregation, the part everyone dreads and nobody adds any value doing, was finished over the course of just an hour.
I got to go to sleep on time that night, which was nice.
One of our Studio members, Darren Smith, asked a fair question in the comments: why Claude Code instead of Cowork? Honest answer is the setup underneath it. I've built a system around Claude Code with .md files, MCP connections, and folder access, and complex jobs that touch four systems at once land better there. If you haven't built that scaffolding yet, Cowork will get you most of the way on a shorter DD list. The gap only shows up when the job needs to reach into your PMS, your file storage, and your team's inbox in the same run.
I probably could have used Cowork, but I just trust Claude Code’s Plan Mode and execution more for these larger projects with a lot of moving parts.
That’s it for this week’s issue. We’d love to hear your thoughts! What you found valuable, what you could do without, or any topics you’d like us to cover in future newsletters. Send us an email and let us know.
If you thought this newsletter was useful, do us a favor & forward it to a colleague who’s curious about AI in CRE. They can subscribe here.
P.S. We’re currently booking consulting engagements & in-person speaking dates on AI in CRE for Q4. If you’re interested in building a custom AI solution for your company, or you’d like us to speak at an upcoming conference or association meeting, head to the appropriate link above and reach out!