Updated 20 September 2026: first published on Elite Agent in March 2026. The article has been refreshed as an AI basics guide to discussing an AI-generated answer with a property client.
Two stories caught my attention in March 2026. In one, a Florida homeowner described using ChatGPT to help organise his sale. In the other, Ryan Serhant described an AI answer unsettling people involved in a major deal.
They raised a useful question for an agent: what do you do when a client arrives with an AI answer that challenges your advice?
Start by looking at it together. A confident answer can unsettle someone even when you have done careful work. It is an opportunity to examine the evidence and explain the reasoning behind the choices.
What the reported examples show
Robert Levine described using ChatGPT to help plan and prepare the sale of his Cooper City home. The original report described the property going under contract within days; that is different from completing settlement.
In a later WUSF interview, published in April 2026, Levine discussed the work he did and the lawyer’s review of the contracts. This was a person using tools and professional help, not a chatbot independently conducting a transaction.
The original video reference is retained as background to the reported example.
Serhant’s account concerned a $50 million deal that he said nearly came undone after the parties separately consulted ChatGPT. The linked source below is his account. We do not have the full prompts, private transaction documents or valuation evidence, so we cannot assess the answers or establish precisely what caused the difficulty.
The examples are reasons to ask better questions. They do not establish that AI caused either outcome or provide a formula for somebody else’s sale.
How do I respond when a client says “ChatGPT disagrees”?
Ask which part of the answer concerns them. With their agreement, look at the question they asked, the information supplied and any cited sources. They do not need to share unrelated private conversations.
Separate four things:
- Facts you can verify, including dates and the basis of comparisons.
- Assumptions the answer makes without establishing them.
- Preferences and constraints the client has actually expressed.
- Missing information that could change the assessment.
An outdated comparable or an invented property feature is a factual problem. A client’s preference for certainty or a particular timetable is a different kind of consideration. Neither should disappear inside a confident paragraph.
Use AI to prepare the discussion
You can ask AI to organise the questions without asking it to choose for the client. Use an approved tool and only information you are authorised to provide.
Help me prepare for a conversation about a property offer using the authorised, de-identified information below.
Verified facts and sources: [DETAILS]
Client’s stated preferences and constraints: [DETAILS]
AI-generated answer they want to discuss: [TEXT]
Missing information: [GAPS]
Separate verified facts, assumptions, preferences and unsupported claims. Check whether the cited material actually supports each point where source material is supplied. Say when you cannot verify something.
Set out options and trade-offs without deciding for the client. Identify evidence that could change the assessment and questions for the relevant agent, valuer, solicitor or other adviser.
Do not invent comparable sales, legal requirements or anyone’s motives. Finish with a plain-English explanation I can check before discussing it with the client.Explain what would change your view
Models can ask questions and acknowledge uncertainty. A fluent response still does not tell you that the important context was supplied. Equally, a human recommendation deserves to be examined rather than accepted simply because a person gave it.
Explain your sources, the assumptions behind your recommendation and what new information would cause you to reconsider. Let the client ask questions and make room for another professional’s input where appropriate.
The useful skill is making the reasoning visible: what you know, what remains uncertain and how the available choices relate to the client’s priorities.
Practise the conversation with your team
Use a fictional scenario in a team session and compare how different explanations sound to the client. If you would like me to run a practical AI session for your team, get in touch about training.
The original version appeared on Samantha McLean’s Substack. Read the refreshed source article on AI Powered Real Estate by Samantha McLean on Substack.