● Learning Platformprototype

Day 8 — AI in Real Estate

Time: ~45 min · Date: Sat May 2

Why this matters

Up to now we've been talking about Claude in the abstract. Today: how this actually shows up in your work. What other agents are doing. What's worth your time. What's worth being skeptical about.

A note on scope: most of the examples below lean residential because that's the most universal vocabulary. But the same patterns apply across residential, commercial, and investment work — just with different inputs. A commercial lease summary works the same as a residential disclosure summary. An investment-property cap-rate analysis works the same as a single-family CMA. Translate freely.

What AI is genuinely good at (in real estate)

Writing the first draft of anything.

  • Listing descriptions
  • Buyer/seller emails
  • Open house ads, social captions, just-listed announcements
  • Disclosure summaries in plain English
  • Follow-up checklists after a showing
  • Listing presentation prep

The pattern: AI writes the first version in 30 seconds, you spend 2 minutes editing. The total is faster than a 15-minute blank-page write.

Summarizing long documents.

  • Inspection reports → 5-bullet summary
  • Disclosure packets → "things a buyer should know"
  • Long client emails → key requests
  • HOA documents → red flags

You still skim the original. AI compresses your skim time.

Comp analysis (assistive, not authoritative).

  • "Given these 3 sales, justify pricing at $X."
  • "What should I emphasize in my CMA?"
  • It's not picking the comps. You are. AI structures the argument.

Brainstorming.

  • "Five ways to position this listing's awkward floor plan."
  • "What questions should I ask this seller in our first meeting?"
  • "What objections should I expect from this buyer?"

Anywhere you'd benefit from a colleague to bounce ideas off, AI is that colleague at 3am.

Personal admin.

  • Drafting weekly to-dos from your notes
  • Summarizing your week's work
  • Creating templates for things you do repeatedly

What AI is not good at (in real estate)

Live data.

  • Current MLS listings (unless you paste them in)
  • Today's mortgage rates
  • This week's market data
  • Real-time anything

If you ask "what's the median home price along the Wasatch Front right now?" Claude is guessing or relying on training data that's months old. Always verify time-sensitive numbers.

Legal advice.

  • Contract interpretation
  • Disclosure requirements
  • Compliance questions

Claude can summarize and explain in plain English, which is genuinely useful. It cannot replace your broker, your attorney, or RESPA. Don't ship legal advice from Claude.

Anything where being confidently wrong matters.

  • Specific commission splits
  • Exact contractual obligations
  • Specific tax implications
  • Lender-specific qualification rules

Claude will sound certain. It might be wrong. The cost of being wrong here is high. Verify.

Personal client knowledge.

  • "What does Sarah care about?" — Claude doesn't know unless you've told it.
  • "What's the right tone with Mark?" — Same.

You're the keeper of relationships. AI is a tool you point at the parts of work that aren't relationship.

Privacy: things to watch

When you paste a client's information into Claude.ai, that conversation goes through Anthropic's servers. They have privacy policies; they don't share with other users. But "their servers see it" is technically true.

Practical advice:

  • For sensitive client info (full names + financials, social security, etc.) — paste structurally, scrub identifiers if you can. "A buyer earning $X with savings of $Y wants to qualify for $Z mortgage" is fine. Their actual name and SSN is not necessary for Claude to do the job.
  • For shared documents — same. Claude doesn't need the seller's full address to help you summarize the inspection.
  • When in doubt, redact what you don't need to share.
  • Once you're using Claude Code on your local Mac (not Claude.ai), the data still flows to Anthropic for processing — but your files stay on your machine. Conversations still travel.

What other agents are doing

A non-exhaustive list of things you might steal:

  1. Daily morning brief. Some agents start the day with: "Read my notes from yesterday. What follow-ups did I commit to? What should I prioritize?"
  2. Listing photo captions for social. Photographer delivers 30 photos. Agent feeds the 5 hero shots in and asks for caption variations.
  3. Showing prep. Before a private showing: "Here's the listing details. Here's what the buyer told me they want. Coach me through the showing — what to highlight, what concerns to address proactively."
  4. End-of-week wrap. "Summarize my work this week from these notes. Pull out wins, follow-ups, things I dropped."
  5. Negotiation prep. "We received this counter-offer. What are the seller's likely motivations? What concessions could I propose that cost us little but feel meaningful?"
  6. Transaction file kickoff. When a deal goes under contract: "Generate my standard transaction checklist for [property type]."

The pattern: AI takes over the parts of your job that are generic professional output — emails, summaries, templates — and frees you to do the parts that aren't (the relationships, the judgment, the negotiation).

Exercise (~25 min)

Write down (on paper, in notes, wherever):

  1. Three tasks you do weekly that you'd love AI to take a first pass at. Be specific. Not "writing emails" — "the post-showing thank-you email I send the same day."
  2. One task you would NOT trust AI with, and why.
  3. One area where you'd want Claude to push back on you — places where a second opinion or counter-argument would actually help. (Example: "When I'm pricing too aggressively to please a seller.")

Bring this list when the laptop arrives. We'll set up some of these as your starter prompts.

Recap

You should now be able to:

  • List 3 tasks where AI is genuinely useful in real estate
  • List 2 tasks where AI is risky or insufficient
  • Describe one privacy practice when pasting client info
  • Imagine yourself using AI in a real workflow — not abstractly

Going deeper (optional, ~45 min)

This space moves fast, so I'm steering you toward search queries rather than specific articles.

Read (~25 min)

  • Google: "AI tools for real estate agents 2024" — skim 2–3 of the top results. Notice which tools are actually doing things versus rebranding existing CRM features.
  • Google: "how realtors use ChatGPT" — case studies and workflow examples. Some will be relevant; some won't.

Watch (~20 min)

Skeptic exercise

Pick one AI tool you encounter that claims to be "for real estate." In your notes, write 3 questions you'd ask the vendor before paying for it. (Hint: "where does my client data go?" should be one of them.)

Done?

  • Day 8 complete

Next: Day 9 — Git: just enough

Lesson recap

Done with this lesson?
Mark it complete to track your progress.
← Day 7
Week 1 Checkpoint
Day 9 →
Git: Just Enough