Tuesdays are 3 AI updates, one hot take, one action. Weekends are 1 practical real estate agent workflow.
On Thursday, Justin Ratowsky and I walked the agents in my Agent as CMO session through how to think about their websites as the hub of how AI finds them and talks about them. Justin is leading the AEO work on my new site through his new company, Auto Authority, so I know he’s putting these recommendations into practice for agents. AEO, short for answer engine optimization, means making your information easy for AI tools to find and repeat correctly. With independent AI agents like Meta's Muse and ChatGPT's dots now running errands for people, it matters more than ever, because those agents go a step past search: they open your site, read it and try to act on it for their user.
What you need. Your screenshots from Tuesday, Claude or ChatGPT with web search turned on, and the login to your website builder. If you use Claude, connect Gmail under Settings, then Connectors, before step 5.
1. Find where AI gets its facts about you (5 min)
Paste this into a new chat:
Search the web for [your full name], real estate agent in [your town]. List every page that describes me: my website, my brokerage profile, Zillow, Realtor.com, Homes.com, my Google Business Profile, LinkedIn and any local press. For each page, copy exactly what it says about my brokerage, years licensed, sales volume or number of deals, price range, neighborhoods, designations like CRS or CLHMS, and languages. Put it in a table and flag every place where two pages disagree. Then tell me how you would answer a buyer who asked, 'Who's the best agent for a [price range] home in [neighborhood]?'
Expect a few mismatches. An old brokerage and an outdated sales total are normal.
2. Pick one version of the truth (5 min)
Using that table, write one fact sheet about me: name, brokerage, the neighborhoods and ZIP codes I actually work, my typical price range, years licensed, designations, languages, and one sales figure I can back up with MLS data, with its time frame (for example, '42 homes sold in 2025'). Add a 50-word bio written for a buyer or seller in [your town], with no 'top' or '#1' claims unless my brokerage's data supports them. Use only facts from the sources you found, and mark anything you're unsure about with [CHECK].
Fix the [CHECK] items yourself, then update the two profiles that disagree most with the fact sheet. AI tools tend to believe what shows up the same way in many places, so matching beats polishing. Only publish sales numbers you can document. Put the fact sheet on your site's About page as plain text too, since that's the page AI tools are most likely to quote about you.
3. Label your website so machines can read it (10 min)
Schema is a small block of hidden code that tells search engines and AI tools who you are in their own format. Ask for yours:
Write schema.org JSON-LD for my real estate website using this fact sheet. Include a RealEstateAgent block with my name, brokerage, phone, email, areaServed listing each neighborhood and ZIP code, knowsLanguage, and sameAs links to my brokerage profile, Zillow, Realtor.com, Google Business Profile, LinkedIn and Instagram. Add a FAQPage block for the four questions buyers and sellers in [your town] ask me most, using my answers: [paste them]. Don't add anything that isn't in the fact sheet or my answers.
Paste the result into your site's header code. Most builders have a setting for it, often called code injection or custom code, and if you can't find it, send the code to whoever manages your site. The FAQ answers also need to appear on the page itself, not only in the code. Then paste your homepage address into the Schema Markup Validator to confirm it reads.
4. Make your site easy for an AI agent to use (5 min)
When a buyer asks Muse or a ChatGPT dot to find them a local agent, it may land on your site and try to pull out how to reach you. Test that yourself:
Read [your website] as if you were an AI assistant working for a family relocating to [your town] with a [price range] budget who wants an agent who knows [neighborhood]. Tell me whether you can find my name, phone, email, the neighborhoods I cover and how to reach me, all in plain text, and whether anything on the site shows I know [neighborhood]. List anything that got in the way, like details that only appear inside images, pop-ups or pages you couldn't open. Then look at [your website]/robots.txt and tell me whether it blocks AI crawlers such as GPTBot, ClaudeBot or PerplexityBot.
Put your phone number and email on the page as plain text, not only inside a graphic or a contact form. If your robots.txt blocks AI crawlers, ask whoever manages your site whether that was on purpose, because an agent that can't read your site will send the buyer to one it can. And if nothing on your site proves you know [neighborhood], that's your next blog post.
5. Close your review gap (5 min)
There’s no certainty, but some experts believe Google reviews count for more in AI search than Zillow reviews, so start there.
Search my Gmail for clients I closed with in the last two years, using closing, settlement and 'congratulations on your new home' emails. Compare that list with the reviews on my Google Business Profile: [link]. For everyone who hasn't left a Google review, draft a short personal email that mentions their home and neighborhood, says a Google review helps the next buyer in [your town] find me, and includes my review link. Save them as drafts in my Gmail and don't send anything.
Ask everyone on the list, not only the clients you know were happy, because Google's review rules don't allow cherry-picking positive reviewers or offering anything in return for a review. Read every draft before it goes.
6. Check again every month
In Claude, turn steps 1 and 4 into a scheduled task for the first Monday of each month, and add this line to it: "Compare against last month's results and tell me only what changed, including any new page that mentions me." It's the same check, without you having to remember it.
Where this breaks
AI answers change from one chat to the next and from one person to the next, so compare patterns across your screenshots instead of reacting to one answer. Schema and matching profiles help AI read you correctly, but they won't move you to the top overnight. I'd expect multiple weeks. Agents like Muse are also new, and what they can do on a website changes month to month, which is why step 4 belongs in the monthly check.
AI FOR RE AGENTS DECEMBER COURSE IS OPEN
My Claude 101 for Real Estate class is tomorrow, Saturday, November 14, three hours live online, and it's the fastest way to get set up for workflows like this one. My end-of-year course runs December 6 to 12 in four live sessions, and the early price ($120 off) ends Sunday. Enroll at maven.com/matt-spangler/ai-for-real-estate. If you buy the December course, Claude 101 is free. Everything else lives at agent-intelligence.co.
“It helped me distill my strategy and messaging for the upcoming year.” - Ronda Prothro, agent in Austin
-Matt
