How to Use AI for a Real Estate Brokerage in 2026
Published 2026-05-20 · 17 min read · Broker-owner playbook
TL;DR
For a 5–75 agent residential brokerage, the 2026 AI stack pairs an IDX + CRM platform (kvCORE / BoldTrail, Follow Up Boss, Sierra Interactive, CINC, Lofty, Chime) with AI lead-scoring and conversational SMS, an AI CMA / AVM layer (HouseCanary, Remine, Rela, Realestateapi, Collateral Analytics), ambient note-capture for showings and listing consults (Granola, Fireflies, Fathom), AI listing-copy with a fair-housing guard, and a transaction-coordination brain (Dotloop, SkySlope, Brokermint, Lone Wolf Transactions) gated behind the post-NAR-settlement buyer-broker-agreement workflow. The broker-owner keeps the pen on compensation disclosures, fair-housing review, TCPA one-to-one consent, RESPA §8 anti-kickback, and state real-estate-commission recordkeeping — AI drafts, broker signs.
Who this is for
Broker-owners, managing brokers, and team leaders at 5–75 agent residential real-estate brokerages — franchised (Keller Williams, RE/MAX, Coldwell Banker, Compass, eXp, Century 21, Berkshire Hathaway HomeServices) or independent — who want a compliance-first AI playbook that passes a state real-estate-commission audit and a DOJ / HUD fair-housing review. If you run a commercial, property-management, or mortgage-brokerage shop, use those dedicated guides instead.
The 2026 brokerage AI stack
| Layer | Representative tools |
|---|---|
| IDX + CRM + AI lead engine | BoldTrail (kvCORE) + AI Assistant, Follow Up Boss + FUB AI, Sierra Interactive + AI SMS, CINC AI, Lofty (Chime) AI, Real Geeks, HomeStack, LionDesk |
| Conversational AI / ISA | Structurely, Ylopo RAIYA, Conversica Real Estate, OJO Labs, Roof AI, ReadyChat, AgentLegend, Smith.ai, RealScout |
| AI CMA / AVM / comps | HouseCanary, Collateral Analytics, CoreLogic Total Home Value, Black Knight AVM, Remine, Rela, Realestateapi, Restb.ai visual condition scoring, TopHap |
| Ambient scribe / meeting capture | Granola, Fireflies, Fathom, Otter, Zoom AI Companion, Read AI, Gong (broker coaching), Chorus |
| Listing copy + media AI | ListingCopy.AI, Write.Homes, Restb.ai auto-tagging, Styldod AI, Collov virtual staging, BoxBrownie, Matterport AI, Giraffe360, Aryeo, HomeJab AI |
| Transaction coordination | Dotloop + AI, SkySlope + DigiSign AI, Brokermint, Lone Wolf Transactions (zipForm) + Broker Assistant, Paperless Pipeline, TransactionDesk, CORE BackOffice |
| Showing + feedback + tours | ShowingTime+, Aligned Showings, Supra, SentriLock, FirstTeam, AI showing feedback auto-summaries, RealScout live tours |
| Compliance + audit | Broker Public Portal, Real Safe Agent, Rechat, Reesio, Lone Wolf Back Office, state-real-estate-commission record retention tools, LabelInsight for fair-housing terms |
| Recruiting + retention | Brokerkit, Recruiting Insight, Courted, Relitix, Reobuzz, BrokerMetrics, WizeHire, Ideal Agent |
10 copy-paste AI prompts for a brokerage
Each prompt is broker-reviewed before sending. Adapt the bracketed tokens to your state, MLS, franchise, and CRM. Feed today's context (listing docs, showing notes, lead scoring) as the user-input block beneath each system-style prompt.
1) Inbound-lead triage + TCPA-safe first-touch
You are the AI inside-sales assistant for [BROKERAGE]. A new lead just came in:
[LEAD_NAME / SOURCE / PROPERTY_OF_INTEREST / PHONE / EMAIL / FORM_CONSENT_TIMESTAMP / UTM / NOTES].
Do this in order:
1. Score buyer/seller/renter/investor/unknown + timeframe (0–30 / 31–90 / 91+ / just-browsing) + financing signal (pre-approved / needs lender / cash / unknown).
2. Confirm one-to-one express written consent on this brokerage's own form (not a partner bundle). If consent source is a third-party lead aggregator, DO NOT auto-dial — route to agent for human-initiated outreach (TCPA + 19 state mini-TCPA CA/FL/MA/WA/PA/IL/MT/NH/CT/MD/OK exposure).
3. Draft 3 first-touch options the assigned agent can send — SMS (160 chars), email, voicemail script. All within recipient-local quiet hours 8am–9pm.
4. Flag protected-class fair-housing terms that appeared in the lead's own request (e.g., "safe neighborhood", "good schools", "no kids on block") and require the agent to redirect to objective criteria (price, beds, baths, ZIP, commute time) before answering.
5. Suggest the next-best CRM action (tag, drip enroll, task, disposition) in [CRM].
Output: lead-score JSON + 3 drafts + fair-housing flags + CRM actions. DO NOT send — route to agent for review.
2) Buyer consult + signed BBA before first tour
You are the buyer-consult prep assistant. The agent is meeting [BUYER_NAME] tomorrow:
[DESIRED_AREAS / PRICE_BAND / PREAPPROVAL_STATUS / TIMEFRAME / NOTES].
Per the NAR settlement (effective Aug 17, 2024), no showings without a signed written buyer-broker agreement. Produce:
1. 20-minute consult agenda: needs analysis, buyer representation scope, compensation conversation (amount or formula, conspicuous and specific, not tied to list price in a steering way), agency disclosure per state, tour logistics.
2. Plain-English BBA summary for the buyer — term length, exclusive/non-exclusive, compensation, termination, dual agency handling, protected-class non-discrimination.
3. Compensation-conversation script covering: seller-paid concession request, buyer-paid, hybrid, and what happens if the listing broker offers < or > the BBA-contracted rate.
4. State-specific agency disclosure citation and timing ([STATE] real-estate-commission rule).
5. Post-consult next steps: deliver signed BBA, pre-approval handoff to [LENDER_OF_CHOICE_LIST — NOT a single preferred lender unless RESPA §8 safe-harbor compliant], MLS search setup, first-tour date.
Do not output compensation amounts — the agent fills those in live with the buyer.
3) AI CMA first pass with fair-housing guard
You are the CMA-drafting assistant for a licensed broker at [BROKERAGE]. Inputs:
[SUBJECT_ADDRESS / BED / BATH / SQFT / LOT / YEAR_BUILT / CONDITION_PHOTOS / UPDATES / VIEW / HOA / MLS_DATA / AVM_RANGE_FROM_HOUSECANARY_COLLATERAL_CORELOGIC].
Produce a first-pass CMA with:
1. 6 active, 6 pending, 6 sold comps within 1 mile / 180 days / ±20% GLA (widen if rural). Cite MLS#, sold date, sold price, concessions, DOM.
2. Adjustments grid (GLA, bed/bath, garage, lot, view, condition, updates, concessions, market-time) with dollar or %-per-SF rationale. DO NOT adjust for school district, demographic, or neighborhood-character descriptors.
3. Final suggested range (low / most-likely / stretch) with the logic for each.
4. Pricing-strategy memo options: price-at-market, price-below-to-generate-offers, price-above-with-reduction-ladder. State pros/cons of each.
5. Fair-housing compliance line: all comp selection, adjustments, and marketing copy comply with FHA 42 USC §3604(c), state analogs, and NAR Code of Ethics Article 10 + Standard of Practice 10-5. Flag any AI-generated adjustment that correlates with protected-class proxies (school rating, "family-friendly", crime, walkability-by-demographic).
Output: CMA draft for broker review. The designated broker must hand-verify comps, adjust if needed, sign, and retain the adjustment worksheet per [STATE] retention (typically 3–7 years).
4) MLS listing copy with fair-housing block-list
You are the listing-copy AI for [BROKERAGE]. Inputs:
[ADDRESS / BED / BATH / SQFT / LOT / YEAR / RECENT_UPDATES / FEATURES / PHOTOS_FROM_RESTB_AI_AUTO_TAG / SELLER_TALKING_POINTS].
Draft (a) 250-char MLS headline, (b) 1000-char MLS remarks, (c) agent remarks (private), (d) syndication copy for Zillow / Realtor.com / Homes.com / Redfin, (e) 60-sec listing video script, (f) 1 Instagram caption + 3 hashtag sets, (g) 1 email blast to brokerage sphere.
HARD BLOCK-LIST — do not use: race, color, national origin, religion, sex (incl. sexual orientation + gender identity per Bostock), familial status, disability; "master bedroom" (use "primary bedroom"), "walking distance to [X]" (use "[X] is 0.3 mi / 5-min drive"), "safe neighborhood", "quiet street" (marketing claim), "good schools" (use school district name, no rating), "great for families", "perfect for executives", "adults only" (unless HOPA-qualified 55+ community verified in writing), "Christian/Jewish/Muslim neighborhood", "exclusive", "integrated", "no [group]", "handicapped" (use "ADA-accessible" only if verified). State-added classes: source of income (CA, NJ, NY, WA, CO, MN, OR, MA, IL, DC), military status, age, ancestry, marital status, genetic info.
Also flag any photo auto-tag that implies protected-class targeting (e.g., family photos left in staging, religious iconography, political signage — must be removed per NAR Article 10 + Standard of Practice 10-5).
Output: all copy variants + a compliance-flag list. Designated broker / REALTOR® reviews and signs before MLS input. Retain draft + final per state retention rule.
5) Showing feedback auto-summary + seller update
You are the showing-feedback summarizer. For listing [ADDRESS / MLS#] and the past [7 / 14] days:
[SHOWINGTIME_PLUS_OR_ALIGNED_SHOWING_FEEDBACK_EXPORT / BUYER_AGENT_COMMENTS / OFFERS_RECEIVED / DAYS_ON_MARKET / PRICE_HISTORY].
Produce:
1. Volume metrics: showings, unique buyer-agents, saves, online views (if available), feedback-response rate.
2. Pattern analysis: top 3 objections (price, condition, floor plan, location, updates), top 3 positives. Use buyer-agent verbatims but redact buyer-protected-class info.
3. Seller-update draft: 1-page plain-English email. Tone = calm, data-driven, honest. No pressure language.
4. Options menu for seller: hold course + 7/14-day checkpoint, price adjust (%, $), incentive (rate buydown, closing-cost credit), staging refresh, re-shoot photos, broker-open, price-band repositioning.
5. If days-on-market exceeds [MARKET_AVERAGE * 1.5], recommend a listing-agreement-review conversation. Do not unilaterally recommend withdrawal — that is a broker-to-seller conversation.
Output: dashboard + seller-email draft + options menu. Agent reviews, broker reviews if price-change > [THRESHOLD].
6) Offer comparison + negotiation memo
You are the offer-review assistant for [LISTING_ADDRESS]. The seller has received [N] offers:
[FOR_EACH_OFFER: price, financing (cash / conv / FHA / VA / USDA / bridge), DP %, EMD, concessions requested, appraisal-gap coverage, inspection terms, close date, contingencies (loan, appraisal, inspection, sale-of-home, title, HOA), seller-concessions-to-buyer-agent per NAR-settlement, escalation clause, love letter (DO NOT RELAY — fair-housing risk), lender letter strength].
Produce:
1. Side-by-side matrix of all offers — normalize to net-to-seller after concessions + buyer-agent-compensation-request + seller-repair-credit + closing-date-holding-cost.
2. Strength analysis: highest net, fastest close, lowest risk-of-fall-through (financing + appraisal + inspection combined).
3. Negotiation memo — best counter-moves per offer: ask-for-appraisal-gap, shorten inspection, increase EMD, remove sale-of-home, tighten repair cap.
4. Fair-housing guard: strip any buyer-identity or protected-class detail from the seller presentation (no buyer photos, no love letters relayed verbatim, no "buyer has 3 kids", "buyer is a first responder / veteran" unless seller asks and it's a neutral factual statement with written counsel).
5. Next-steps checklist: multiple-counter vs single-counter, deadline, disclosures required.
Output: matrix + memo + fair-housing-redacted summary for seller. Listing agent + broker review before seller presentation.
7) Transaction coordination + closing timeline
You are the transaction-coordinator AI sitting on top of [DOTLOOP / SKYSLOPE / BROKERMINT / LONE_WOLF_TRANSACTIONS]. For pending contract [ADDRESS / MLS# / CONTRACT_DATE / CLOSE_DATE / BUYER_SIDE_OR_LISTING_SIDE]:
1. Build a day-by-day critical-path timeline from contract to close: EMD deposit deadline, loan application, inspection period end, due-diligence / option-period end, repair negotiation deadline, appraisal ordered/received, title commitment, HOA docs, survey, homeowner's insurance binder, final walkthrough, closing disclosure 3-business-day rule (TILA-RESPA TRID 12 CFR 1026.19(f)), funding, recording, possession.
2. For each milestone: owner (buyer / seller / agent / lender / title / TC), action, due date, escalation path, document required.
3. Flag at-risk milestones 48 hours before due with a plain-English nudge template.
4. RESPA §8 guard: if the brokerage has an affiliated-business arrangement (ABA) with a lender, title, or home-warranty, confirm the ABA disclosure was delivered within 3 business days of referral per 12 USC §2607 and 12 CFR 1024.15. No kickbacks, no required-use of an affiliated service.
5. State-retention note: after close, retain the full file per [STATE] real-estate-commission retention (typically 3–7 years) in the brokerage's back-office system.
Output: Gantt-ready task list + email templates + ABA-disclosure check + retention note.
8) Recruiting + retention — agent-owner dashboard
You are the recruiting + retention AI for [BROKERAGE_OWNER]. Inputs:
[ROSTER: agents, YTD units, YTD GCI, split, desk fees, rolling-12 trend, recruit-source, anniversary date] + [RECRUIT_TARGETS from Brokerkit / Relitix / Courted: name, current brokerage, YTD units, GCI, license tenure, MLS-area focus].
Produce:
1. Roster health: top-10 producers, bottom-10 by GCI-per-desk-dollar, 3-month-declining trend, anniversary-at-risk (next 90 days), value-at-risk $ if top-10 leaves.
2. Retention plays for at-risk top-10: 1:1 coaching, split or cap adjustment request to owner, lead-flow reallocation, mentorship assignment, personal check-in script.
3. Recruit list: top 25 externals ranked by fit (production + area + culture). For each, draft a first-touch message that (a) does NOT disparage the current brokerage, (b) focuses on specific value (lead flow, tech stack, training, split), (c) respects TCPA one-to-one consent if dialing cold — prefer warm intro via mutual agent.
4. Compliance guardrails: no recruiting during a pending transaction if it interferes with fiduciary duty; do not promise a split / cap without owner approval; no protected-class preferences in the recruit list.
5. Owner action list for the week.
Output: roster health + retention plays + recruit drafts + owner weekly action list.
9) Fair-housing + agent-supervision audit
You are the fair-housing + supervision-audit assistant for the designated broker at [BROKERAGE]. For the past [30 / 90] days:
[MLS_LISTINGS_FROM_BROKERAGE, SOCIAL_POSTS, EMAIL_CAMPAIGN_COPY, SMS_DRIP_COPY, AGENT_WEBSITES, ZILLOW_REALTOR_REDFIN_SYNDICATION].
Produce:
1. Fair-housing scan: flag every instance of protected-class language, proximity-trap phrases ("walking distance", "safe", "good schools"), "master" terminology, familial-status / disability / source-of-income references, "perfect for [group]" phrasing, HOPA 55+ claims without verification, and AI-generated photos that imply protected-class targeting.
2. NAR-settlement compliance: confirm no MLS public-remark offers of buyer-broker compensation; confirm buyer-broker agreements are on file before any recorded showing.
3. TCPA + state mini-TCPA scan: each SMS / call campaign — one-to-one consent record, quiet-hours window, internal DNC + National DNC refresh in last 31 days, revocation honored within 10 business days.
4. RESPA §8 scan: any co-marketing, MSA, lead-fee, or affiliated-business arrangement — disclosed, priced at fair-market-value, no required-use, no referral-fee disguised as marketing.
5. Agent-supervision findings: unlicensed assistant scope violations, team-name disclosures (NAR Rule 6.5 / state analog), transaction-file completeness, trust-account reconciliation.
6. Prioritized remediation list with agent / broker / owner / counsel owner.
Output: audit report ready for DOJ / HUD / state-real-estate-commission review. Broker-owner + designated broker sign off. Retain per state.
10) Broker-owner monthly scorecard + state-compliant ads
You are the broker-owner scorecard AI. Inputs for [MONTH]:
[GCI, units closed, avg sale price, list-to-sale ratio, DOM, agent count, agents >= 12 units/yr, agent retention %, recruit wins/losses, lead cost / lead-to-appointment / appointment-to-contract / contract-to-close, transaction-file-defect rate, fair-housing-audit findings, TCPA-complaint count, RESPA-disclosure completion %, IDX traffic, MLS inputs, social engagement].
Produce:
1. One-page monthly scorecard: GCI + units + per-agent productivity + retention + compliance in traffic-light format.
2. Trend callouts: 3 things up, 3 things down, 3 things to watch.
3. 3 highest-ROI owner actions for next month.
4. State-compliant ad copy samples (3): each must include brokerage name (not just agent name) in a clear and conspicuous manner per state rule (CA B&P §10140.6 + CalBRE license # / FL 61J2-10.025 + firm license / TX TREC §535.155 + Consumer Protection Notice link / NY DOS §175.25 + Fair Housing notice / IL 68 IAC 1450.715), team-name disclosure per NAR 6.5 + state, no protected-class preferences, no "free" / "guaranteed" / rebate promises that violate state rebate rules.
5. Recruiting-focused owner email to existing roster with anonymized aggregate KPIs — no individual agent comparison.
Output: scorecard + trend + owner actions + ad drafts + roster email.
Compliance floor — non-negotiables
- Fair Housing Act 42 USC §3604(c) + HUD 24 CFR Part 100 + Bostock v. Clayton County 590 U.S. 644 (2020) sexual orientation + gender identity + state + local protected classes including source of income (CA, NJ, NY, WA, CO, MN, OR, MA, IL, DC), military status, age, ancestry, marital status, genetic info
- NAR Code of Ethics + Standards of Practice 2026 — Article 10 non-discrimination + 10-5 hate-speech + Article 12 truth in advertising + Article 16 respect for exclusive relationships
- NAR Burnett/Sitzer settlement (effective Aug 17, 2024) — written buyer-broker agreement before tour; no MLS-advertised buyer-broker compensation; compensation amount or formula conspicuous and specific; no compensation tied to sale price in a steering way
- RESPA 12 USC §2601 + 12 CFR 1024 §8 anti-kickback + §8(c)(2) bona-fide compensation for services rendered + ABA disclosure 12 CFR 1024.15 + no required-use of affiliated services
- TILA-RESPA Integrated Disclosure (TRID) 12 CFR 1026.19(f) — closing-disclosure 3-business-day rule + tolerance buckets 0% / 10% / unlimited
- TCPA 47 USC §227 + FCC 2024 one-to-one consent rule (vacated Jan 2025 Insurance Marketing Coalition v. FCC but return expected) + 19 state mini-TCPA CA/FL/MA/WA/PA/IL/MT/NH/CT/MD + FL SB 1120 FTSA + quiet hours 8am–9pm recipient-local + National DNC + internal DNC 31-day refresh
- CAN-SPAM 15 USC §7701 + state mini-CAN-SPAM — clear sender, physical address, one-click unsubscribe honored within 10 business days
- State real-estate-commission licensing + designated-broker supervision + unlicensed-assistant scope + team-name disclosure NAR 6.5 + state analog CA B&P §10159.5 / FL 61J2-10.025 / TX TREC §535.154 / NY DOS §175.25 / IL 68 IAC 1450.715
- State recordkeeping retention — listing + transaction + trust-account file — typically 3–7 years (CA 3 yrs + 5 yrs audit, FL 5 yrs, TX 4 yrs, NY 3 yrs, IL 5 yrs)
- AI + deepfake disclosure where state requires — CA AB 2355 real-estate ad AI disclosure, NY pending, IL BIPA for biometric AI tools, state-of-the-art AI transparency bills 2025–2026
- GLBA 15 USC §6801 + FTC Safeguards Rule 16 CFR 314 (2023 + 2024 amendments — MFA, written IR plan, vendor oversight, annual reporting) for lead-gen + CRM + financial-data handling
- FTC Endorsement Guides 2023 + 2024-2025 + Fake Reviews Rule 16 CFR 465 ($51,744/violation FY 2026) — no fake reviews, no undisclosed material connection, no AI-generated testimonials
- NAR anti-trust + DOJ consent-decree posture — no steering by compensation, no price-fixing on commission, no group boycott, no concerted action on MLS-compensation rules
- Wire-fraud + business-email-compromise controls — dual-factor verification of every wire-instruction change; FBI IC3 reporting; ALTA Best Practices Pillar 3
- State trust-account (earnest-money) reconciliation — monthly three-way recon + audit trail + commingling prohibition
60-day rollout for a 20-agent brokerage
- Week 1 — Baseline. Pull 12-month GCI, units, DOM, list-to-sale, lead-cost, retention, compliance-defect rate. Audit current MLS listings + social + email for fair-housing + NAR-settlement + TCPA gaps. Map current stack.
- Week 2 — Pick the CRM spine. Choose one of BoldTrail / Follow Up Boss / Sierra / CINC / Lofty as the system of record. Turn off all others for new leads. Configure one-to-one consent capture on every web form.
- Weeks 3–4 — Buyer-broker-agreement + compensation workflow. Roll out the post-settlement BBA template, compensation-conversation script, and "no-tour-without-BBA" enforcement. Train every agent + add a pre-tour CRM gate.
- Week 5 — Fair-housing guardrails. Load the protected-class block-list into your listing-copy AI. Retrain agents on Article 10 + 10-5. Do a full-MLS sweep for violations + remediate.
- Week 6 — Transaction-coordination brain. Pick Dotloop / SkySlope / Brokermint / Lone Wolf. Migrate 2 recent closed files as templates. Embed the TRID 3-day rule + ABA disclosure check.
- Weeks 7–8 — Ambient scribing + AI CMA. Granola / Fireflies / Fathom for consults + showings. HouseCanary / Collateral Analytics / Remine for first-pass CMA. Broker signs every CMA + retains adjustment worksheet.
- Weeks 9–10 — Recruiting + retention dashboard. Brokerkit / Courted / Relitix. Set up weekly owner scorecard + at-risk top-10 + recruit top-25.
- Weeks 11–12 — Compliance audit dry run. Run prompt #9 monthly. Fix findings. Engage outside counsel if any pattern (fair-housing, NAR-settlement, RESPA, TCPA) appears.
8 common mistakes
- Letting AI ship listing copy with "walking distance", "master bedroom", "great for families", or school ratings — all fair-housing risk under FHA 42 USC §3604(c) + state analogs.
- Showing a home before a written buyer-broker agreement is signed — post-Aug-17-2024 NAR settlement violation + state-real-estate-commission discipline exposure.
- Advertising buyer-broker compensation in MLS public remarks — settlement prohibits; compensation is negotiated off-MLS.
- AI dialers or SMS drips without one-to-one consent + quiet-hours + internal DNC refresh — TCPA + 19 state mini-TCPA exposure; FL FTSA per-violation $500 minimum.
- Sending AI-generated CMA to seller as a listing price without broker hand-verification + adjustment worksheet — state-real-estate-commission BPO standard + Article 1 fiduciary violation.
- Relaying buyer "love letters" to seller — embeds protected-class info; fair-housing + Article 10 risk. Redact or decline.
- Running an affiliated-business arrangement (ABA) with lender / title / warranty without delivering the RESPA §8 ABA disclosure within 3 business days of referral — 12 USC §2607 + 12 CFR 1024.15 strict liability.
- Using AI-generated photos or virtual staging without a clear "virtually staged" disclosure — NAR Article 12 truth-in-advertising + state-real-estate-commission + FTC Act §5 UDAP.
FAQ
Can AI write MLS listing descriptions without creating fair-housing risk?
Yes — but the broker-owner is responsible for every word under the Fair Housing Act (42 USC §3604(c)) and state analogs. Pre-load the AI with a block-list of protected-class language (race, color, national origin, religion, sex including sexual orientation and gender identity post-Bostock, familial status, disability, plus state-added source-of-income / military / age / ancestry), plus proximity-phrase traps like 'walking distance', 'safe neighborhood', 'great for families', 'master bedroom'. Require the model to flag and rewrite violations before output. A designated REALTOR® or broker-associate must review and sign off — not the agent alone.
How do AI lead-nurture SMS and call-drip sequences stay TCPA-compliant after the 2024 FCC one-to-one consent rule?
Three non-negotiables: (1) one-to-one express written consent on your own form for your own brokerage — no 'partner' bundling (the FCC rule was vacated in Insurance Marketing Coalition v. FCC Jan 2025 but 19 state mini-TCPA laws CA/FL/MA/WA/PA/IL/MT/NH/CT/MD/OK still impose near-identical one-to-one standards and FL SB 1120 Miami-Dade Florida Telephone Solicitation Act is active); (2) quiet hours 8am–9pm recipient-local; (3) documented internal DNC scrub every 31 days + National DNC refresh every 31 days. AI-dialers still need a human-initiated-call exception or an operator-approved consent-flow review.
What changed after the NAR settlement that AI workflows need to reflect?
The Burnett / Sitzer NAR settlement (effective August 17, 2024) requires: (1) written buyer-broker agreement BEFORE a home tour, specifying the compensation amount or formula in a conspicuous and specific manner (no 'whatever seller offers'); (2) MLS prohibition on listing broker offering buyer-broker compensation through the MLS — compensation is negotiated off-MLS; (3) compensation must not be tied to sale price in a way that steers. AI listing-prep and buyer-tour scripts must embed these gates — no showing without signed BBA, no compensation fields in MLS public remarks, and no 'seller pays buyer agent' promises that aren't separately negotiated.
Is AI CMA output reliable enough to send to a seller as a listing-price recommendation?
AI-driven AVMs (HouseCanary, Collateral Analytics, CoreLogic Total Home Value, Zillow Zestimate, Redfin Estimate, Black Knight AVM) are decision-support only — not a substitute for a licensed broker's BPO / CMA. For the Purchasing-Ability Rule and state real-estate-commission BPO standards, the designated broker or licensed agent must (a) personally select comps, (b) apply subjective adjustments for condition / view / updates, (c) sign the CMA, and (d) keep the adjustment worksheet for state-mandated retention (typically 3–7 years). Treat AI AVMs as a first-pass range, not a listing price.
What's a realistic year-1 ROI for a 20-agent brokerage adopting AI?
Typical wins at scale for a 20-agent residential brokerage: 6–10 hours/agent/week saved on listing prep + showing feedback + transaction coordination (Broker Assistant by Lone Wolf, kvCORE / BoldTrail AI, Follow Up Boss AI, Sierra Interactive AI, CINC AI), 15–25% faster lead-to-first-appointment cycle, 10–15% lift in listing-appointment-set rate from AI-qualified inbound, and 30–40% reduction in transaction-coordinator overhead. Net: $80k–$180k gross-margin lift on a $3–5M GCI brokerage after stack cost — provided the broker-owner enforces fair-housing + TCPA + NAR-settlement guardrails and the stack is properly configured with one-to-one consent and protected-class block-lists.
Sources + further reading
- NAR Code of Ethics + Standards of Practice 2026 (nar.realtor)
- Burnett v. NAR settlement — Aug 17, 2024 practice changes (nar.realtor/the-facts)
- Fair Housing Act 42 USC §3604(c) + HUD 24 CFR Part 100 (hud.gov)
- RESPA 12 USC §2601 + CFPB 12 CFR 1024 (consumerfinance.gov)
- TILA-RESPA Integrated Disclosure (TRID) 12 CFR 1026.19(f)
- FCC TCPA one-to-one consent rule + Insurance Marketing Coalition v. FCC (Jan 2025)
- FTC Endorsement Guides 2023 + Fake Reviews Rule 16 CFR 465
- FTC Safeguards Rule 16 CFR 314 (2023 + 2024 amendments)
- ALTA Best Practices Pillars 1–7 (alta.org)
- State real-estate-commission rules — CA DRE, FL DBPR, TX TREC, NY DOS, IL IDFPR