When I first met AI, ChatGPT 3.5, it felt like magic. Here was this tool, super friendly, always hyped me up and agreed with me; it felt like a charming new imaginary friend. It was useful for me at the time because it read through and explained my divorce paperwork.
It would also make suggestions and draft legal documents; this is where I discovered what kind of magic it performed; it didn’t replace my attorneys, but it did help me to provide the information the attorneys wanted. And as with all magic, it must first be practiced and then always used with caution.
How to use AI for disclosures
Here are the magic spells I use step-by-step with AI on disclosures.
Real estate: Domus revelio
I started using an AI to help with buyer disclosure packages because I was overwhelmed. I was working for municipal government and tapering my real estate business down. The city property inventory wasn’t going to manage itself, and neither were my buyers’ disclosure packages.
I had been using AI for months to review legal documents, leases, and policy manuals; I had graduated from ChatGPT to Google Gemini because Gemini has a huge amount of “context” — a measure of how much information an assistant can take in and effectively work with.
After some incredible blunders, then lots of practice, I figured out a way to adopt the magic.
How it works: Expecto extractum
Here’s the process that works, refined through hard experience:
First, I scan the entire disclosure package myself. The greatest hits and the natural hazards disclosure (NHD) cover. I’m not just looking for red flags in the standard forms (SPQ, TDS, AVID); I’m specifically hunting for the addendums, the supplemental letters, the extra pages that are non-contiguous to the main documents. These are where sellers often bury their most critical admissions, separate from the checkboxes.
Then, I feed the entire digital package into an AI like Gemini for a deep dive, cross-document analysis.
Gap: Lumos extractus
The next step is the most important: I conduct a gap analysis. I compare the AI’s report against the list of concerns I flagged during my manual read. This is where I consistently find the AI’s blind spots.
Has it missed the handwritten note in the margin of the TDS about “occasional dampness in far closet”? Possibly it overlooked the scanned, cursive letter from the seller explaining a past repair.
When I find these gaps, I re-prompt. I direct the AI to specific pages and sections, asking if it can decipher the handwritten note or the supplemental PDF as a primary source. Some models simply cannot decipher physical handwriting, whether cursive or print, treating it as an unreadable image rather than actionable data.
Missing these details is a rookie move, one I’ve made a few times as a rookie and as a vet — see the “incredible blunders” comment above. Remember it’s no fun being called out on an obvious mistake that our experience should prevent.
Damage: Obscuro fractura
The National Association of Realtors’ settlement — buyers having to figure out their own representation if the seller won’t pay — has put a spotlight on the necessity for complete seller disclosures. The financial stakes are real. Remediating mold and water damage averages $2,000 to $10,000, with structural repairs costing far more. This is the post-closing surprise that torches client trust and leads straight to arbitration.
In a cooling market where sellers downplay flaws, and with California courts scrutinizing material defects, a superficial review is how lawsuits happen. The California Association of Realtors constantly updates what a “material fact” is, putting the burden on us to use judgment, not just checkboxes.
Most agents skim packages, glance at reports and move on. That’s how you get expensive surprises — water damage worse than expected or title problems hiding in plain sight.
Users: Nuntius inspexit
NAR’s data shows 32 percent of agents haven’t used AI in their business yet, while only 20 percent use it daily. The gap isn’t technology — it’s understanding. Agents either avoid AI entirely or trust the output blindly without verifying what it missed.
The 20 percent using AI daily aren’t winning because they’ve handed everything over to ChatGPT. They’re winning because they’ve figured out where AI adds value and where human judgment stays irreplaceable.
They use AI to synthesize technical data — inspection reports, title documents, financial statements — then apply their own expertise to the parts AI can’t handle: seller admissions, handwritten notes, supplemental letters that contain the real story.
The prompt: Veritas revelio
The prompt I use to have an AI review disclosure package:
“Act as a real estate agent. Please analyze this entire disclosure package thoroughly, identify all material issues and seller admissions, cross-reference details across documents, and flag any risks or inconsistencies needing agent attention.”
Gemini costs $20 a month. Your ability costs nothing but time. Invest both and you save hours, avoid headaches, and protect your clients and yourself.
America Foy is Chief Real Estate and Development Officer at Where Ever, Inc. Connect with him on LinkedIn and Instagram.
