Home · Real estate · AI automations
AI for real estate agents: 16 real examples from agents, teams and brokers
Updated October 18, 2026 · 16 sourced examples · no paid placement
Short answer: Agents and brokers mostly use AI on the paperwork and follow-up around a deal: sweeping the CRM for leads that went quiet, matching buyer inquiries to new listings, drafting listing descriptions and pricing presentations, writing a data-backed price-reduction email, planning showing days, and turning how the office works into written SOPs. In these 16 examples the AI is mostly Claude or ChatGPT; three teams also use an AI voice agent for calls they miss. Reported results include listing descriptions drafted in minutes instead of 45 minutes to hours, about 40 hours a month saved by one Realtor, a blog post cut from five or six hours to under an hour at a brokerage, and a seller agreeing to a $100,000 price reduction after an AI-drafted email. These are what each agent reported, not promises.
Below are 16 documented examples from solo agents, teams, brokerages and a property manager, told by the agents themselves or reported by Inman, REALTOR News, HousingWire, Business Insider and others: who did it, what the AI does, what they reported and where the story came from. After that: what to keep human, what to watch out for and how to try one this week.
Other industries too: see real AI automations by task for any small business.
What these 16 businesses actually did
- A morning CRM sweep that drafts follow-ups: William Zhang (solo agent, Austin) (owner-told, via William Zhang, Real Estate AI Society blog). A Claude-based assistant pulls Follow Up Boss leads, notes and deals, scores each lead A–D using the Keller Williams model, drafts a personalized follow-up, and creates a dated task. He sends the text himself. Reported: One buried lead it flagged turned into a deal (self-reported). Setup took real time, and a person still clicks send. Full case
- Every buyer inquiry matched to new listings: Georgie Smigel (Coldwell Banker team, Pennsylvania) (press, via Pittsburgh Post-Gazette via Denver Gazette (Tribune News Service)). Every inquiry (price range, neighbourhoods, school districts) goes into a database her AI tools search; she asks who is looking for a given price and area and invites them to showings. AI also drafts her ads. Reported: The team was on track for about $100 million in sales, roughly $10 million more than in 2024, with AI credited and no graphic designer needed. Full case
- Listing copy, client follow-ups and market reports: Missy Plunkett (South Carolina and Tennessee) (press, via Accessible AI newsletter). ChatGPT drafts listing descriptions, personalised follow-ups using past conversations with each client, and neighbourhood market reports from data she uploads. Reported: Self-reported: a listing description takes about 20 minutes instead of hours, about 40 hours a month saved overall, and client response rates up almost 30%. Full case
- Pricing with seven prompts and presentations in Gamma: Chuck Bachteler (Berkshire Hathaway, Florida) (press, via Real Estate AI Flash podcast, EP 086). He downloads property, neighbourhood and market reports from RPR, uploads them to ChatGPT or Perplexity, runs seven prompts in order to reach a price, then turns the output into a presentation in Gamma. Reported: Self-reported: listing descriptions that took 45 minutes to an hour now take minutes, and presentations come together 'within minutes'. He stresses checking the output because AI 'can lie with confidence'. Full case
- A CMA turned into a price-reduction email, plus a team voice doc: Driscoll Peca team (Ottawa area) (owner-told, via Real Estate AI Flash EP 123: How Jaime Peca Built AI Without a Tech Background (Rajeev Sajja)). She pulled the listing's MLS sheet, ran a new CMA on 30-day sold data, and gave both to an AI assistant along with the hard conversation she needed to have; it drafted a detailed email explaining the numbers. She also wrote a brand-voice document (how the team speaks, what it does and doesn't say, no emoji blasts or em dashes) that every agent loads into the AI tool they use. Reported: She says the email would have taken her at least an hour to put together, and after she sent it the seller agreed to a $100,000 price reduction. She now keeps an hour every Thursday for agents to bring their AI questions. Self-reported. Full case
- A farming list of high-equity owners and MLS descriptions: Camila Castillo (Compass, Coral Gables) (press, via REALTOR News (NAR): This AI tool is moving the needle for brokers). She asks Claude Cowork to make a spreadsheet of properties bought seven or more years ago, then has it check the county property appraiser site to see whether each owner's mailing address matches the property address. A saved Skill writes MLS descriptions: she gives it the address, the executed listing agreement and the last MLS tax roll. She also runs a weekly call teaching her colleagues to use it. Reported: The farming research that 'would take us weeks' is handled by the agent, and MLS descriptions that took 45+ minutes each are now drafted automatically from three inputs. The article stresses that agents still review and verify the output. Full case
- Listing descriptions in his own voice: Randy Baruh (Corcoran, New York City) (press, via Business Insider). The broker uses the free ChatGPT to draft property listings and other mundane writing, then edits for accuracy and SEO. Reported: The broker describes saving time on mundane tasks; edits are still required. Full case
- Listing proposals and offering memorandums as Claude skills: PCI (Cleveland investment sales team) (owner-told, via Real Estate Pros Show (Investor Fuel): AI for Real Estate Agents: How Tal Tamir Automates His Brokerage With Claude). After bringing in a consultant at the start of the year, he says he has lived in Claude five days a week for three to four months, creating skills and workflows including new listing proposals. The team is building an OM skill linked to its CRM: you tell it which property, it pulls the photos and details and produces the OM. Reported: He calls the OM skill a real time saver, but it was still being built when he spoke and he gives no figures. The CRM isn't named. Full case
- Showing instructions and a tour route from a list of homes: William Zhang (owner-told, via Real Estate AI Society: AI MLS Automation — Look Up Showing Instructions and Schedule Tours Hands-Free (William Zhang)). He gives Claude Code the MLS listing IDs (or addresses). Claude uses computer-use browser control to open Unlock MLS, pull showing instructions, lock-box type (Supra vs combo), contact info, owner/tenant status and notes, then filters out expired, withdrawn, under-contract and income-restricted listings that do not fit the buyer. It builds a Google Maps route for vacant go-and-show properties and a readiness checklist for which homes need agent outreach for codes. He saves the flow as a reusable Claude Skill so the next list is one command. Reported: Self-reported: the MLS lookup that used to take about 30 minutes by hand ran about 15 minutes unattended while he stepped away; browser automation is slower than an API but he does none of the clicking. No closed-deal or GCI figure is tied only to this workflow. Full case
- SOPs from screen recordings and a back office on Claude: Bozeman Real Estate Group (Montana) (owner-told, via Inman: How we rebuilt our brokerage's back office with AI workflows). Over three months they moved every process into Notion and set up a Claude project for each department. Staff record themselves doing a task, a transcript-to-SOP skill turns the recording into steps and checklists, and an SOP-to-Notion skill publishes it with the required fields. New hires ask Claude how things are done. Blog posts go from an admin's outline through a fact-checker skill and then an SEO/GEO optimizer skill, and every content piece is checked against brand guidelines. Reported: Self-reported: a blog post went from five to six hours to under an hour, and an SOP now takes minutes instead of the hours she expected. Her caveat: it was a leadership-level rebuild (they also moved their server to Dropbox), not a quick fix. Full case
- A brokerage dashboard watched by AI agents: Bramlett Partners (Texas) (press, via REALTOR News (NAR): This AI tool is moving the needle for brokers). Using Claude Opus for agentic coding (after earlier 'vibe coding' in Replit), Bramlett and five developers built a brokerage management platform with one dashboard and 17 AI agents, each with a single job, such as monitoring emails or phone calls or checking for compliance. He can ask it how the business is doing; it reviews every area and flags what's broken. The agents surface issues; he reviews them and makes the decisions. Reported: About 100 hours over roughly three months, versus an estimated $2 million before; he says agentic coding lets them build 10–15x faster. In one check-in it reported email deliverability and revenue looked fine but one process was broken. He is now using Claude to build a CRM, the last non-proprietary piece of his stack. Full case
- Tenant maintenance triaged by text: Nulf Management Services (Pittsburgh) (press, via Pittsburgh Post-Gazette via Denver Gazette (Tribune News Service)). His maintenance system uses AI to troubleshoot tenant issues by text before assigning the right technician. His maintenance director turns walk-through notes into room-by-room bid lists, and an AI bookkeeping tool reads utility bills. Reported: It saves the hours staff spent triaging tickets. He also used ChatGPT to draft a sales agreement when buying two firms; check legal documents with a lawyer. Full case
- A month of social posts, then repurposing the winners: Josh Ries (lead-gen coach, 30-day test) (owner-told, via Inman, by Josh Ries). Each evening ChatGPT drafted 20 to 30 short Threads posts, which he scheduled; the best performers became Canva quote cards, then 30 to 60 second videos, then longer YouTube videos. Reported: About 110,000 views across Threads, TikTok, Meta and YouTube in 30 days (self-reported). Full case
- Buyer calls qualified by a voice agent that hands off when unsure: LKW Home Team (Columbus, Ohio) (press, via Real Estate News (Keller Williams sponsored): Keller Williams-affiliated agents share top AI plays). A voice AI agent answers inbound buyer calls and, in under two minutes, collects name, phone, email, ideal neighborhood, price range, where they are in the search and whether they're pre-approved. It offers to send listings and connect them with a lender. It's trained on the team's SOPs, school and safety ratings, compliance rules and market updates, and transfers to a human whenever it's less than 95% confident. Web-form leads become tagged CRM contacts with a voice-note booking prompt; open-house sign-ins use a QR form that triggers follow-up with consent. Separately, ChatGPT drafts and posts social content from topics kept in a Google Sheet. Reported: the agent has answered 900+ inbound calls and taken 400+ next-step actions; the owners say automation frees 10–20 hours a week, and the three-agent team grew GCI 35.8% going into 2024. The voice platform isn't named. Full case
- A safety net for missed lead calls, plus AI call coaching: RE/MAX Advantage (Nashville) (press, via HousingWire: How RE/MAX Advantage uses AI for calls and agent training). Speculo's AI agent 'Remi' picks up inbound lead calls the team misses, answers listing questions like beds, baths and square footage, and transfers live to a Realtor when the caller is urgent or asks something an unlicensed assistant shouldn't answer. It also re-engages the existing database. A second tool, Shilo, grades agents' calls, writes call summaries and action items, and runs AI role-play; new agents must do at least two role-play calls a week. Reported: Ashton calls Remi 'our safety net if we happen to miss a call,' and Beagle says 'we have seen an increase in agent performance' from the call grading and role-play. No specific numbers given. Full case
- A phone assistant that books tours and screens spam: William Zhang (owner-told, via Real Estate AI Society (William Zhang): I built an AI voice agent to never miss a real estate lead again). When he doesn't pick up, a voice agent named Ashley answers within two rings. A first step sorts the caller into buyer/seller lead, general inquiry (like another agent's showing request) or spam. Leads go to a scheduling path that checks his calendar, offers two time slots at a time and books the appointment with name and email. Showing requests and other inquiries become a detailed message with up to two follow-up questions (it doesn't give out lockbox or gate codes; it relays the request to him). Spam calls are ended immediately with no notification. Reported: Self-reported: he checks the call log once a day and shows up to booked appointments; he estimates the showing-request calls alone used to take close to an hour a week. The examples in his post are test calls; no lead or deal counts shared. He puts typical voice-agent costs at $50–$200 a month depending on volume. Full case
These are each company's own results with their own tools and teams, as reported by the source. Vendor case studies are written by the company selling the tool. None of them are promises.
What do real estate agents automate with AI?
Follow-up. William Zhang, a solo agent in Austin, has a Claude-based assistant pull his Follow Up Boss leads each morning, score them, draft a personal follow-up and create a dated task; he sends the text himself, and says one buried lead it flagged became a deal. Georgie Smigel's Coldwell Banker team puts every buyer inquiry into a database her AI tools search, so she can ask who is looking at a given price and area and invite them to showings. Missy Plunkett has ChatGPT draft follow-ups using past conversations with each client and reports client response rates up almost 30%.
Listings and pricing. Plunkett says a listing description takes about 20 minutes instead of hours. Chuck Bachteler uploads RPR reports to ChatGPT or Perplexity, runs seven prompts in order to reach a price and builds the presentation in Gamma. Camila Castillo has a saved Claude skill draft MLS descriptions from the address, the listing agreement and the tax roll, and had Claude Cowork build a farming list of owners who bought seven or more years ago. Jaime Peca gave an AI assistant a fresh CMA and the listing sheet to draft a calm, data-backed price email; she says the seller then agreed to a $100,000 reduction.
The back office. Kate Hulbert's brokerage in Bozeman records staff doing a task and has Claude turn the recording into an SOP in Notion; a blog post went from five or six hours to under an hour. Eric Bramlett built a brokerage dashboard where 17 single-job AI agents watch email, calls and compliance and flag what is broken. Zhang has Claude Code pull showing instructions from the MLS and plan the tour route, about 15 unattended minutes instead of 30 of clicking.
Calls are one example among many. LKW Home Team, RE/MAX Advantage and Zhang use AI voice agents for calls nobody picks up, each set to hand off to a person when the caller needs one.
What should stay human?
Sending, pricing and advice. Zhang's assistant drafts, but he clicks send. Bachteler checks every output because AI "can lie with confidence". REALTOR News stresses that agents still review and verify what Castillo's agent produces. LKW's voice agent transfers to a person whenever it is less than 95% confident, and RE/MAX Advantage's hands urgent callers and questions an unlicensed assistant shouldn't answer to a Realtor.
Legal documents. Dustin Nulf used ChatGPT to draft a sales agreement when buying two firms; have a lawyer check anything like that.
What to watch out for
Setup is real work. Zhang says his CRM assistant took real time to set up, and Hulbert calls her brokerage's rebuild a leadership-level project over three months, not a quick fix.
Who is telling the story. The LKW Home Team article is sponsored by Keller Williams, and every result above is self-reported by the agent or broker.
Your voice. Jaime Peca's team wrote one brand-voice document (how they speak, what they don't say) that every agent loads into the AI tool they use, so drafts sound like the team.
Tools named in these examples
| Tool | Starting price (base plan) | Checked |
|---|---|---|
| ChatGPT | Free $0; Go $8/mo; Plus $20/mo; Pro $100/mo | 2026-10-17 |
| Claude | $17/mo billed annually ($20 monthly) | 2026-10-18 |
| Perplexity | Pro $17/mo billed annually (promo shown on Pro page) | 2026-10-18 |
| Gamma | Plus $9/seat/mo billed annually ($108/yr); Pro $18/seat/mo billed annually | 2026-10-18 |
| Canva AI (Magic Studio) | Pro $18/mo or $144/yr for one person | 2026-10-18 |
| ElevenLabs | $6/mo (Starter); free plan available | 2026-10-15 |
Starting prices from each vendor's pricing page on the date shown. AI phone and texting features can be priced separately from the base plan, so check before you buy. Also named in these examples: Follow Up Boss, RPR, Notion, Unlock MLS, Twilio and TidyCal (prices not checked here).
How to start this week
- Pick the one job that eats your week outside showings and clients: following up with old leads, listing descriptions, pricing presentations, or showing-day prep.
- Gather what AI needs to draft from: your CRM export or lead notes, a past listing you like, your CMA or RPR reports, and a short note on how you talk to clients.
- Paste those into Claude or ChatGPT with one clear ask (a follow-up text for each quiet lead, an MLS description, a price email to a seller) and edit the first drafts yourself.
- Check every number, address and claim against the MLS or the report, and send it from your own phone or email.
- Once the drafts are reliably close, make it routine: a saved Claude skill or ChatGPT project, a morning CRM sweep, or an automation inside the CRM you already pay for.
Have Claude sweep your CRM each morning and draft follow-ups for leads you'd forget
A Claude-based assistant pulls Follow Up Boss leads, notes and deals, scores each lead A–D using the Keller Williams model, drafts a pers…
ClaudeFUFollow Up BossMatch every buyer inquiry to new listings by asking AI
Every inquiry (price range, neighbourhoods, school districts) goes into a database her AI tools search; she asks who is looking for a giv…
via Pittsburgh Post-Gazette via Denver Gazette (Tribune News Service) · 2026-01-03Use ChatGPT for listing copy, client follow-ups and market reports
ChatGPT drafts listing descriptions, personalised follow-ups using past conversations with each client, and neighbourhood market reports …
ChatGPT
Canva AI (Magic Studio)Price listings with a seven-prompt ChatGPT routine and build presentations in Gamma
He downloads property, neighbourhood and market reports from RPR, uploads them to ChatGPT or Perplexity, runs seven prompts in order to r…
ChatGPT
Perplexity
GammaTurn a fresh CMA into a calm, data-backed price-reduction email for sellers, and give every agent a shared brand-voice doc
She pulled the listing's MLS sheet, ran a new CMA on 30-day sold data, and gave both to an AI assistant along with the hard conversation …
Quoted: Jaime Peca, team lead, Driscoll Peca team (The Real Brokerage) · via Real Estate AI Flash EP 123: How Jaime Peca Built AI Without a Tech Background (Rajeev Sajja) · 2026-05-20Use an AI desktop agent to build a farming list of high-equity homeowners and draft MLS listing descriptions
She asks Claude Cowork to make a spreadsheet of properties bought seven or more years ago, then has it check the county property appraise…
ClaudeCCClaude CoworkWrite listing descriptions with ChatGPT and keep the agent's voice
The broker uses the free ChatGPT to draft property listings and other mundane writing, then edits for accuracy and SEO.
ChatGPTBuild Claude skills for listing proposals and offering memorandums that pull from your CRM
After bringing in a consultant at the start of the year, he says he has lived in Claude five days a week for three to four months, creati…
ClaudeHave Claude open the MLS, pull showing instructions and plan the tour route so a property list becomes a showing day hands-free
He gives Claude Code the MLS listing IDs (or addresses).
ClaudeUMUnlock MLSGMGoogle MapsTurn recorded walkthroughs into SOPs and run a brokerage's back office on Claude projects
Over three months they moved every process into Notion and set up a Claude project for each department.
ClaudeNNotionDDropboxVibe-code a brokerage dashboard where AI agents watch email, calls and compliance and tell the owner what needs attention
Using Claude Opus for agentic coding (after earlier 'vibe coding' in Replit), Bramlett and five developers built a brokerage management p…
ClaudeCCClaude CodeRReplitTroubleshoot tenant maintenance issues by text with AI before sending a technician
His maintenance system uses AI to troubleshoot tenant issues by text before assigning the right technician.
ChatGPTDraft a month of real estate social posts with ChatGPT and repurpose the winners
Each evening ChatGPT drafted 20 to 30 short Threads posts, which he scheduled; the best performers became Canva quote cards, then 30 to 6…
ChatGPTCanva AI (Magic Studio)TThreadsHave an AI voice agent take a real estate team's inbound buyer calls, qualify them and hand off when it isn't sure
A voice AI agent answers inbound buyer calls and, in under two minutes, collects name, phone, email, ideal neighborhood, price range, whe…
ChatGPTVAVoice AI agent (not named)CNCRM (not named)Use an AI voice agent as a safety net for missed real estate lead calls and AI role-play to train agents
Speculo's AI agent 'Remi' picks up inbound lead calls the team misses, answers listing questions like beds, baths and square footage, and…
Build a solo agent's AI phone assistant that books buyer tours, logs showing requests and hangs up on spam
When he doesn't pick up, a voice agent named Ashley answers within two rings.
ElevenLabsTTwilioTTidyCalQuestions
How are real estate agents using AI?
In these examples: drafting follow-ups to quiet leads from the CRM (William Zhang, Missy Plunkett), matching buyer inquiries to new listings (Georgie Smigel), listing descriptions (Plunkett, Randy Baruh, Camila Castillo), pricing and listing presentations (Chuck Bachteler, Jaime Peca, PCI), showing-day prep (Zhang), SOPs and back-office systems (Bozeman Real Estate Group, Bramlett Partners), social posts (Josh Ries) and answering missed calls (LKW Home Team, RE/MAX Advantage, Zhang).
What are the best AI tools for realtors?
There isn't one best tool. The agents here mostly use Claude (Zhang, Castillo, Hulbert, Bramlett, PCI) or ChatGPT (Plunkett, Baruh, Bachteler, Ries), with Perplexity, Gamma and Canva for research and presentations and an AI voice agent for missed calls. Start with the job you want off your plate, then pick the tool.
Can ChatGPT write listing descriptions?
Yes, as a first draft. Plunkett says a listing description takes about 20 minutes instead of hours, Bachteler says 45 minutes to an hour became minutes, and Baruh edits each draft for accuracy and SEO. Check every detail against the MLS before it goes live.
Can AI do a CMA or pricing for a listing?
It can work from the data you give it. Bachteler uploads RPR reports and runs seven prompts in order to reach a price, and Peca gave an AI assistant a fresh CMA on 30-day sold data to draft a price-reduction email. Both start from real reports, and Bachteler checks everything because AI can be confidently wrong.
Can someone set it up for us?
Yes. The $197 Quick-Win sets up one automation in accounts you own, live within 48 hours of getting access, with a refund if it is not live in 48 hours or is not doing its job in the first 14 days.
Related guides
- AI automations for any small business, by task
- All real estate AI examples
- AI for proposals, quotes and estimates: real examples
- AI client intake and onboarding automation: real examples
- AI email triage and inbox management: real examples
- Small businesses using AI agents: every example by industry
- Have one automation set up for you ($197)
One short email a day: 3 new AI use cases from real small businesses, each linked to its source, plus a Tool of the Day. Free, no spam, unsubscribe anytime.
Want one set up for you?
The $197 Quick-Win: we set up one AI automation that saves you hours each week, in accounts you own, live within 48 hours of getting access. No call needed. Refund if it isn't live in 48 hours or isn't doing its job in the first 14 days.
See the $197 Quick-Win Bigger setups