AI on the Record · evidence register

Real cases, not predictions · First industry: real estate

Real cases of real estate agents using AI, each one linked to its source.

One person, one task, one tool, in their own words. We rewrite each case in plain words, keep the number the person states, mark it self-reported, say what they sell, and link to the video.

Read the overview reportOpen the case register

No tool reviews, no prompts to copy, no vendor case studies. Nothing is invented and nothing is verified.

  • 27casesone video, one task each
  • 20peopleagents and broker-owners
  • 6tasks5 with a page of their own
  • 2platforms11 YouTube · 16 TikTok
  • 13cases with a business tielabelled on the case
  • 8cases that share a numberall self-reported

Enter by task

What agents use AI for

One page per task. Each page has a summary, the cases, where it goes wrong, and the people who argue against it.

  • Live

    Listing descriptions

    Agents paste their notes and the property facts into an AI tool, get a draft, and edit it into their own voice.

    6 cases · 5 people1 argues against2 have a business tie

  • Live

    Social media and local market content

    Agents give an AI tool local data (an MLS export, two neighbourhood names, or a Canva template) and get back a post, a video script or a newsletter, which they edit and publish.

    4 cases · 3 people1 has a business tie

  • Live

    Lead follow-up and CRM

    Agents connect an AI assistant to their CRM (here, always Follow Up Boss) and let it read the notes, rank who to call today, and type in new contacts for them.

    4 cases · 2 peopleAll 4 have a business tie

  • Live

    Transaction management

    Once a deal is moving, agents hand the repeat paperwork to AI: one prompt after signing turns into a closing checklist, a buyer email and calendar dates, Claude drafts an offer, and a team leader has bots sort and file email across 30–50 active deals.

    4 cases · 4 people1 has a business tie

  • Live

    CMA and comps

    Agents let Claude collect the comps and build the CMA report, then they check the comps and set the price themselves.

    5 cases · 3 people4 have a business tie

  • Waiting: 3 sources

    Other tasks

    AI cold calling, an agent bio, an AI testimonial video and team business plans: one record each.

    4 cases · 4 people1 has a business tie

How we work

Four rules, applied to every case

  1. A source link on every case

    Each case links to the YouTube or TikTok video where the person says it, with the time in the video.

  2. Results are self-reported

    We keep the number the person states and mark it self-reported. We do not verify it and we do not adjust it.

  3. Business ties are labelled

    When the person sells a course, a kit, a paid community or a service, the label says what. 13 of 27 cases carry one.

  4. Screenshots are credited

    Each screenshot names the person, the platform and the time in the video. We do not use stock photos.

Read the full method: who counts, labels, and the order of cases →

Enter by tool

The two tools agents name most

Tool page

Claude for real estate agents

Named in 13 cases by 9 people: listing descriptions, social media and local market content, lead follow-up and CRM, transaction management and CMA and comps.

Open the Claude page →

Tool page

How to use ChatGPT for realtors

Named in 9 cases by 7 people: listing descriptions, social media and local market content and other tasks. Custom GPTs count as ChatGPT.

Open the ChatGPT page →