An agent’s work comes in waves tied to the deal: prepare for the listing appointment, market the property, keep a contract on schedule, close, then stay in touch. Each stage has its own writing and organizing load, and that is where AI chat tools help. This playbook goes stage by stage, with a prompt for each and a note on what the agent has to check.
Stage 1: Before the listing appointment
Agents usually walk in with comparable sales and a pricing view. AI can turn the agent’s own notes on the comps into a clear one-page narrative for the seller.
I'm a real estate agent preparing for a listing appointment. Here are my notes on the subject property and five comparable sales (addresses removed):
[SUBJECT: beds, baths, square feet, lot, condition, upgrades]
[COMPS: sold price, date, beds/baths/sq ft, key differences]
Write a one-page pricing narrative for the seller in plain English: what the comps show, how the subject compares, and a suggested price range with the reasoning.
Use only the figures I've given. Flag anything that looks inconsistent.The agent sets the price. The tool only explains the reasoning the agent has already done.
Stage 2: Marketing the listing
Listing copy gets its own guide on this shelf, with more fair-housing detail. The short version: describe the property, not the people who might live there. Social captions, open-house emails and flyer text follow the same rule.
Write three social media captions (under 60 words each) for an open house.
Property facts: [BEDS, BATHS, FEATURES, OPEN HOUSE DATE AND TIME]
Describe the home and its features only. No statements about who the home is ideal for.Stage 3: Under contract
Once an offer is accepted, the work becomes deadlines and coordination. A timeline table is easy to generate from the contract dates.
Contract accepted on [DATE]. Contract terms:
- Inspection period: [X] days
- Appraisal contingency: [X] days
- Financing contingency: [X] days
- Closing date: [DATE]
Build a timeline table with each deadline, the calendar date, who is responsible, and a reminder date three days earlier.
Note that I need to confirm how days are counted under the contract and local practice.How days are counted (calendar or business days, and what happens on weekends) depends on the contract form and state practice. The agent confirms that, not the tool.
Stage 4: Closing week
Clients appreciate a plain checklist of what happens at closing and what to bring. A generic version can be drafted in advance and tailored per deal.
Write a closing-week checklist for a first-time buyer in [STATE]: what to expect each day, documents to bring, how funds are usually sent, and a warning about wire fraud asking them to confirm any wiring instructions by phone with the title company.
Keep it under 250 words.Stage 5: After closing
Follow-up messages at one month, one year and at each home-purchase anniversary keep an agent in touch with past clients. AI can draft a small library of them in one sitting, written in the agent’s voice once the agent supplies a sample of their own writing.
Here are three emails I've written to past clients, so you can match my voice:
[PASTE SAMPLES WITH NAMES REMOVED]
Draft four short follow-up emails for a buyer client: one month after closing, six months, one year, and a seasonal home-maintenance reminder.
Each under 100 words, no sales pitch in the first two, and a placeholder for one personal detail I'll add myself.The placeholder matters. A message with one real detail the agent remembers reads very differently from a template, and the tool can’t supply that detail.
What to keep the tool away from
Some parts of a transaction shouldn’t be drafted by a chatbot at all. Contract terms, contingency language and disclosure forms come from the brokerage’s approved forms and, where needed, a real estate attorney. Advice on loan programs belongs with the lender. Pricing decisions belong with the agent, using MLS data the agent has pulled and checked. A chat tool doesn’t have live MLS access unless the brokerage has connected it, so any market figure it offers without a source should be treated as unverified.
The checks every stage shares
| Check | Why |
|---|---|
| Facts match the agent’s records | The tool fills gaps with plausible guesses |
| No statements about the kind of buyer or tenant | Fair Housing Act advertising rules |
| Claims are accurate and current | Truthful advertising duties |
| No client financial details in the prompt | Confidentiality and fiduciary duties |
NAR’s Code of Ethics (effective January 1, 2026), Article 12, requires REALTORS® to be honest and truthful and to “present a true picture” in advertising and other representations. That applies to AI-drafted copy as fully as to anything else.
Don’t paste this in
Clients’ financial details, pre-approval letters, the seller’s reasons for selling or bottom-line price, other buyers’ offer terms, lockbox and alarm codes, or anything a client told the agent in confidence. Prompts and outputs should also avoid describing who a home suits by race, color, religion, sex, disability, familial status or national origin. Under 24 CFR 100.75, notices, statements and ads indicating a preference or limitation on those grounds are unlawful, and that includes the words and images used. Wording should describe the property, not the buyer.
Agents running their own business books can borrow the sequence in AI for bookkeeping and accounting, which starts with procedures and leaves financial data until the tools are approved.