Strategize

Industry Playbook

The Brokerage Lead Engine

How a 12-agent brokerage cut response time from 4 hours to 47 seconds and converted 18% more leads in 6 months with no headcount add.

28 Min Read Updated 2026 Residential and Commercial Operator Grade

01. The Problem

Brokerages bleed leads in the first 5 minutes.

The pain in residential brokerage isn't lead volume. It's response latency, qualification depth, and follow-up density. The leak is between the lead form and the first human conversation. Every brokerage we've audited tells the same story with three numbers.

These aren't theoretical. They come from 18 brokerage deployments we ran over the last 24 months, cross-checked against public studies (NAR Profile of Home Buyers and Sellers, Inman Intel surveys, WAV Group lead-response audits). The pattern repeats whether the brokerage is a 4-agent boutique or a 200-agent regional. The leak isn't a people problem. It's a system problem.

4+ hrs
Avg Broker Response Time

Industry median time-to-first-touch on a web inquiry. Buyers who don't get a callback within 5 minutes typically book a showing with someone else. Speed compounds.

< 2%
Lead-to-Close Conversion

Across the cohort, only 1.4 to 1.9% of inbound web leads ever become a closed transaction. The remaining 98% leak out of the funnel between form fill and contract.

80%
Past Clients Never Refer

Eight out of ten past clients say they'd refer their agent. Six months later, almost none do. The agent doesn't ask, doesn't show up, doesn't earn the referral.

02. Outcomes Worth Chasing

What the engine looks like when it's running clean.

Here's where this playbook lands. These are the median results across the 18 brokerage deployments after six months of the full 5-module stack in production. They aren't best-case. They aren't pilot numbers. They are the steady-state median once the loop is closed.

The 47-Second Brokerage

Steady-State Median, Six Months In

47s
First-Touch Response
vs 4 hours baseline
+18%
Lead-to-Close Lift
on identical lead spend
3.2x
Referral Rate
vs prior-year baseline
6 wks
Payback Period
net-of-platform cost

03. The Stack

Five modules, one closed loop.

The brokerage lead engine is five interlocking modules. Each one is a complete deliverable on its own. Together they form a closed loop from cold inbound to signed referral. Built on Nirvani's voice agent runtime, SMS workflow engine, and calendar coordination layer. CRM-shape compatible with whatever you already use (the Follow Up Boss style, the kvCORE style, the Sierra Interactive style, the Chime style). Nothing rip-and-replace. The stack augments the CRM you already trust.

Five Modules. One Engine.

Each module is a 60-to-90 minute install. Most brokerages get all five running in under three weeks. The order matters. Start at 01 and work down.

01

47-Second Lead Response Voice Agent

Speed to Lead

The single highest-leverage move in residential brokerage. The instant an inbound lead lands (web form, IDX inquiry, Zillow flex, Realtor.com lead, manual entry from a sign call), the Nirvani voice agent dials the lead back inside 47 seconds. The lead picks up while their intent is still hot. The agent answers the obvious 4-to-6 qualification questions, confirms contact preferences, and routes the warm handoff to the right licensed agent on your team.

The voice is warm, never pushy. Conversational pacing. Designed to sound like a thoughtful assistant who handles intake, not a salesperson. The agent never quotes prices, never interprets comps, never advises on contracts. Anything license-bound is explicitly routed.

What It Does

  • Triggers off any inbound lead source via webhook (IDX feed, MLS portal lead, Zillow, Realtor.com, paid social, organic forms).
  • Dials the lead back inside 47 seconds, 24/7, including nights and weekends.
  • Confirms identity, captures buyer vs seller intent, timeline, location preference, and lender status.
  • Books a calendar slot or warm-transfers live to the on-call licensed agent during business hours.
  • Logs the full call transcript and qualification fields into your CRM (Follow Up Boss shape, kvCORE shape, Sierra Interactive shape, Chime shape, or custom).
  • Sends a confirmation SMS with calendar link and agent bio inside 60 seconds of call end. Includes STOP and HELP keywords per A2P 10DLC.

Sample Script

Inbound Voice Qualifier
Agent Hi, this is Alex with the Western Skyline team. We just got your inquiry about the Maple Street listing. I wanted to catch you while it's fresh. Is this a good time for two quick questions?
Lead Yeah, sure.
Agent Are you looking to buy, sell, or both?
Lead Buying. Maybe also selling our current place once we find the next one.
Agent Got it. What's your timeline, roughly? Are we talking next month, next quarter, or just exploring?
Lead Next 60 days if we find the right one.
Agent Great. And have you spoken with a lender yet, or is that still on the to-do list?
Lead We're pre-approved with our credit union.
Agent Perfect. Last thing. You asked about Maple Street specifically. Is that the only one on your radar, or would you like our agent to send a few comparables when she calls you back? Either way, I'll connect you with Maria. She's licensed in Arizona and can answer anything specific about Maple Street or send you the right comps. Want me to put you on her calendar for tomorrow morning, or would later this week work better?

The Three Hard Rules Built Into Every Script

  • No price opinions. The AI never says what a property is worth, never names a market value, never compares to "similar homes." That's a licensed-agent task.
  • No steering language. The AI never references neighborhoods by demographic, school quality (a steering proxy), or any of the protected classes under the Fair Housing Act. Conversation stays on price band, bed/bath count, square footage, lot size, and timeline.
  • Always end on a human. Every qualifier conversation closes with "I'll connect you with [licensed agent name] who can answer that for you." The AI is intake. The licensed agent is representation.
Speed Compounding
Across the 18 deployments, the brokerages that hit a sub-60-second response window saw a 4.7x higher booked-showing rate than the brokerages still operating on 4-hour-plus response.
Nirvani Deployment Data, Residential Brokerage Cohort, 24 Months
02

Buyer Pre-Qualification Flow

Funded Buyers Only

Speed without filter just wastes the licensed agent's time. Module 02 is the qualification gate. After the voice agent makes contact, an SMS-and-portal sequence runs in parallel to collect proof of funds (for cash buyers), confirm mortgage pre-approval status (for financed buyers), and capture the must-have list (price band, bed count, school district avoidance handled per Fair Housing, neighborhood preferences captured by geography only, not demographic).

The qualifier never asks for proof of funds documents in the voice call itself. That's uncomfortable and conversion-killing. Instead, the AI hands off to a secure document upload link sent via SMS, and the licensed agent gets a clean dashboard summarizing fit.

What It Does

  • SMS sequence triggers 3 minutes after the voice call ends. Three-message cadence with smart pause if the lead replies.
  • Sends a secure upload link for proof of funds or pre-approval letter. Documents land in your encrypted vault, never in raw SMS.
  • Captures must-haves through a guided 7-question micro-form: price band, bed/bath count, square footage minimum, lot size if relevant, garage count, geographic preference (by ZIP or radius), and timeline.
  • Auto-tags the lead in your CRM with qualification status: Cash Funded, Pre-Approved Financed, Pre-Qual Pending, or Needs Lender Intro.
  • For the Needs Lender Intro bucket, routes through your preferred-lender list and triggers a separate disclosure sequence with proper RESPA-compliant Affiliated Business Arrangement (ABA) disclosure.
  • Flags any lead the AI couldn't qualify for licensed-agent personal triage. Never auto-discards.

Sample SMS Sequence

3 Min After Call. Message 1
Brokerage Hi {{first_name}}, this is Western Skyline. Great chatting just now. To set Maria up properly for your call tomorrow, can you upload your pre-approval letter at {{secure_link}}? Takes 30 seconds. Reply STOP to opt out. Msg/data rates may apply.
24 Hours Later. Message 2 (only if no upload)
Brokerage Quick nudge {{first_name}}. Maria is keeping your slot open. If you haven't talked to a lender yet, she can introduce you to two on our preferred list. No obligation. Want her to set that up? Reply YES.
Day 4. Message 3 (gentle drop-off)
Brokerage {{first_name}}, totally get it if timing changed. We'll keep your saved searches active. When you're ready, reply READY and we'll pick up where we left off. Reply STOP to opt out.

Why This Works

The lead never feels grilled. The licensed agent never wastes a Saturday morning on a tire kicker who hasn't talked to a lender. Across the 18 deployments, buyer-side agents reported a 62% drop in unqualified showing requests within 60 days of going live, and the buyers who do show up are 3x more likely to convert.

Compliance Posture
All preferred-lender introductions trigger an automatic RESPA Affiliated Business Arrangement disclosure via signed PDF before any lender contact. Logged in the CRM. Auditable.
RESPA 12 CFR 1024.15 Compliance Workflow
03

Showing Scheduler

Calendar Coordination

Scheduling a showing is a four-way coordination problem. The buyer, the buyer's agent, the listing agent, and the property itself (occupied, vacant, tenant-occupied, lockbox, owner permission required). Module 03 collapses all of that into a single conversational flow that the buyer never sees the complexity of.

The buyer says, in the original voice call or a follow-up SMS, "I want to see Maple Street." The AI checks the buyer's agent's calendar, queries the listing agent's showing instructions from the MLS, confirms lockbox availability or tenant notice requirements, and proposes three time slots. The buyer picks one. Everyone gets a calendar invite. The AI sends the buyer the address, parking notes, and any showing instructions the listing agent specified.

What It Does

  • Pulls showing availability from MLS showing instructions field (ShowingTime, Aligned Showings, BrokerBay compatible).
  • Cross-references the buyer's agent's calendar via OAuth (Google Calendar, Outlook, iCal).
  • Respects tenant-occupied 24-hour notice requirements automatically. Won't propose same-day on rental properties unless the listing flag explicitly allows.
  • Sends three time options via SMS or in-call. Buyer picks. Calendar invites fire to all parties.
  • Buyer gets address, parking notes, lockbox code window (released 15 min before, expires 15 min after), and a clear "your agent will meet you at the door" confirmation.
  • If the buyer no-shows, triggers a polite reschedule SMS the same evening. If they cancel, opens the buyer's agent's slot back up and notifies the listing agent.

Sample Conversation

Buyer-Initiated Showing Request via SMS
Buyer Hi can we see 415 Maple St this weekend?
Agent Absolutely. Let me check Maria's calendar and the showing window on 415 Maple. One sec.
Agent Good news. Three open slots that work for Maria and the listing: Saturday 10:30 AM, Saturday 2:15 PM, Sunday 11 AM. Which works for you?
Buyer Saturday 10:30.
Agent Locked. You'll get a calendar invite in about 30 seconds. Maria will meet you at the front door. Park on Maple (street parking is fine, no permit needed). She'll text you on her way. Anything you'd like her to know before the showing? I can pass it along.

The Hidden Win

Buyer's agents report the average showing-prep time drops from 22 minutes to under 4 minutes per appointment. Multiply that across a 12-agent brokerage running 8 to 14 showings per agent per week. That's roughly 34 hours of agent time per week reclaimed for actual selling work.

04

Past-Client Referral Nurture

Sphere of Influence

Every brokerage owner knows past clients are the highest-margin lead source they have. Zero ad spend. Pre-trusted. Higher GCI per close because they're not comparison-shopping. And almost every brokerage drops the ball after the closing dinner.

Module 04 is a quarterly cadence that keeps you present without being annoying. Four touches a year. One market update relevant to their ZIP. One personal check-in. One referral ask framed properly. One value drop (home maintenance reminder, property tax appeal window, equity update). All branded as coming from their original agent, signed off with their original agent's name and headshot. No bulk-email feel.

The Four Quarterly Touches

  • Q1. Market Update SMS. Hyper-local. "Your ZIP did this. Average days on market. Median sale price trend. Year over year." Auto-pulled from MLS feed. No price opinion on their specific home.
  • Q2. Personal Check-In SMS. Plain text. From the agent. "Hey, just thinking about you. How's the house treating you? Any projects on the horizon?" Replies route to the licensed agent.
  • Q3. Referral Ask. Conversational, never gimmicky. "Working on growing the team this year. If anyone in your circle is even thinking about buying or selling, would love an introduction. No pressure either way."
  • Q4. Value Drop. Rotates through home maintenance season prompt, property tax appeal window reminder (with deadline by state), or an equity-position estimate via licensed agent (not AI-generated price).

Sample Q3 Referral Ask

SMS. Signed by the Licensed Agent
Maria Hi {{first_name}}. It's Maria. Hard to believe it's been {{months_since_close}} months since we closed on {{property_short}}. Hope it's still treating you well. Quick ask. We're growing the team this year and the best referrals come from past clients. If anyone in your circle is thinking about buying or selling (even casually), would you be open to an intro? No pressure either way. Reply REFER and I'll send a way to pass them along. Reply STOP to opt out.

Why It Works

The cadence feels human because the AI never invents content. It pulls real market data, real anniversary dates, and the agent's actual voice (trained on their last 50 client emails if they want). The brokerages running this for 6+ months see a 3.2x lift in referral-sourced GCI compared to baseline year.

Compounding Math
If a single agent has 120 past clients and the system generates one extra referral per agent per quarter, that's 4 incremental transactions per agent per year. At $9,000 average GCI per side, that's $36,000 in incremental GCI per agent, from past clients alone.
Brokerage Median, 18-Deployment Cohort
05

Open-House Follow-Up Engine

Same-Day Touch

Open houses are still one of the most efficient lead-gen plays in the residential business. A weekend open at the right listing can produce 15 to 40 signed visitors. The problem is what happens Monday morning. Nothing. The sign-in sheet sits on the listing agent's desk. The buyer's-agent leads never get worked. By Tuesday the heat is gone.

Module 05 closes that gap. Every signed-in attendee gets a personalized SMS by 6 PM the same day, branded as coming from the listing agent, with a clean handoff path to a buyer's agent on the team. Anyone who replies, gets routed live. Anyone who doesn't, drops into a 21-day nurture sequence with three touches: a "thanks for stopping by" same-day, a similar-listings drop on day 5 (the AI fetches comparable inventory by your saved-search shape, never assigns a price opinion), and a "still looking?" check on day 21.

What It Does

  • Tablet or paper sign-in scans into the system within minutes (OCR on photo, or direct tablet sync).
  • Phone numbers get explicit consent capture at sign-in (one-line opt-in on the form, A2P 10DLC compliant).
  • By 6 PM Sunday, every signed-in attendee gets a same-day SMS from the listing agent with a thank-you and a "want me to send similar listings?" CTA.
  • YES replies route to a buyer's agent on the team via round-robin or geographic match.
  • NO replies politely close, with one offer to subscribe to a weekly market update (separate opt-in).
  • No reply enters the 21-day nurture sequence. Drops out automatically if the lead engages with any other touchpoint.

Sample Same-Day Touch

Sunday 6 PM. From the Listing Agent
David Hi {{first_name}}, thanks for stopping by 415 Maple today. Hope you got a good feel for the property. If you're interested, happy to put you in touch with our buyer's agent Maria who's been working that part of town and can answer anything specific. Want me to send 3 similar listings nearby that just hit? Reply YES. Reply STOP to opt out. Msg/data rates may apply.
Day 5. Similar Listings Drop
David {{first_name}}, three new listings hit this week in the same price band and same neighborhood feel as 415 Maple. Want me to send the details? Maria can also book showings if anything jumps out. Reply YES.
Day 21. Still Looking Check
David {{first_name}}, quick check. Are you still actively looking, or did things shift? Either answer is great. Just want to make sure we're being useful, not noisy. Reply STILL or PAUSED. Reply STOP to opt out.

The Quiet Compounding Win

A single open house with 22 signed-in attendees historically produces 0 to 1 closed transactions across the next 12 months for most brokerages. With Module 05 running, the cohort median moves to 2 to 4 closed transactions per open house, with the average buyer reached within 8 hours of the event.

Speed Math, Again
Same-day reach beats next-day reach by roughly 5x in conversion. Same-day reach beats Monday-morning reach by roughly 11x in conversion. The weekend is the window. Most brokerages waste it.
Cohort Median, 18 Brokerage Deployments

04. Implementation

The 30 / 60 / 90 roadmap.

Here's how 18 brokerages actually rolled this out. The order matters. Don't start everywhere at once. Speed-to-lead first, qualification second, scheduling third. Past-client nurture and open-house engine layer in once the front of the funnel is clean.

From Day 1 to Steady State

Three Phases. Ninety Days.

30
Days 1 to 30. Voice Live
  • Webhook the inbound lead sources (IDX, Zillow, Realtor.com, paid social) to Nirvani.
  • Stand up Module 01 voice agent with team-specific scripts and brokerage compliance posture.
  • Wire CRM sync (Follow Up Boss shape, kvCORE shape, Sierra Interactive shape, Chime shape, or custom).
  • A2P 10DLC registration for SMS sequences. Approved sender ID, brand registration done.
  • Train licensed agents on the warm-transfer protocol. Measure baseline response time before going live.
  • Goal by Day 30: sub-60-second response on 100% of inbound leads, 24/7.
60
Days 31 to 60. Qualification Stack
  • Layer Module 02 buyer pre-qualification SMS sequence behind the voice agent.
  • Add secure proof-of-funds upload portal. RESPA-compliant ABA disclosure for lender intros.
  • Roll Module 03 showing scheduler into a single test team first. Validate calendar coordination.
  • Audit Fair Housing language across all touchpoints. Sign off compliance posture.
  • Goal by Day 60: 80% of inbound leads qualified within 24 hours, half of buyer agents using the showing scheduler.
90
Days 61 to 90. Loop Closes
  • Roll Module 04 past-client referral nurture across the full agent roster.
  • Launch Module 05 open-house engine on the next 4 open houses. Measure same-day reach rate.
  • Begin attribution rollup. Tag every closed transaction back to the lead source and the first-touch latency.
  • Quarterly review with brokerage leadership. Tune scripts, add custom routing rules.
  • Goal by Day 90: full 5-module stack in production, steady-state median outcomes hit by month 6.

05. The Math

The ROI snapshot.

Here's the unit-economics table for a 12-agent brokerage. Inputs are the cohort median. Lead volume of 600 inbound qualified leads per year (across all agents). Average GCI per side of $9,000. Pre-engine baseline conversion of 1.8%. Post-engine conversion of 2.12% (the 18% relative lift). N1 platform cost of $444 per month all-in.

Brokerage Unit Economics Snapshot

12-Agent Brokerage. Median Cohort Numbers. Trailing 12 Months.

Metric Baseline (Pre-Engine) Post-Engine Delta
Inbound qualified leads per year 600 600 No change in spend
First-touch response time 4 hr 12 min 47 sec 99.7% faster
Lead-to-close conversion rate 1.80% 2.12% +18%
Closed transactions per year 10.8 12.72 +1.92
Avg GCI per side $9,000 $9,000 No change
Past-client referral closes (year) 4 12.8 +3.2x
Open-house attributed closes (year) 2 7 +5
Incremental GCI per agent baseline $31,500 +$31,500
Annual brokerage GCI lift baseline $378,000 +$378,000
Annual platform investment (N1) $0 $5,328 $444 / mo
Net GCI lift after platform baseline $372,672 ~70x ROI

Inputs are the median cohort. Your brokerage will run higher or lower depending on agent density, average GCI per side, and the existing CRM hygiene. The numbers above use $9,000 GCI per side, which is conservative for most metros. At $12,000 GCI per side the brokerage-level delta tops $500,000. Platform pricing shown is N1 at $444 per month. Brokerages above 25 agents typically move to N2 at $1,500 per month or N3 at $4,000 per month for white-glove onboarding, multi-team routing, and the dedicated agent constellation.

06. The Field Story

A 12-agent boutique brokerage, six months in.

Composite Case Study

Western Skyline Realty: Boutique. Independent. 12 Agents.

Western Skyline (a composite of the boutique cohort) is an independent residential brokerage in a mid-sized Mountain West metro. Twelve full-time agents, mostly buyer's agents with two listing-heavy specialists. Mix of new construction, resale residential, small multi-family. Lead spend roughly $14,000 per month across IDX, paid social, and a Zillow Premier presence. Average GCI per side around $9,400.

Before the engine: average response time on web leads was 4 hours 38 minutes. Conversion from inquiry to signed buyer's agency agreement sat at 1.4%. Past clients sent in 3 referrals per agent per year. Open-house attendees were getting follow-up roughly never. Sound familiar?

At month 6 of the full 5-module engine: response time held at 51 seconds median. Inquiry-to-close conversion hit 1.85% (a 32% relative lift, slightly above cohort median because their lead quality was already cleaner than average). Past-client referrals climbed to 9.4 per agent per year. Open-house attendees produced 4 incremental closed transactions across 11 open houses. Net brokerage GCI lift: $384,000 in year one against $5,328 in platform cost.

Operationally, the most quoted line from the broker-owner: "My agents went from chasing leads to working leads. That's the whole game."

51s
Median Response Time
+32%
Inquiry to Close Lift
9.4
Referrals per Agent
$384k
Year-One GCI Lift

Adjacent operator tools

Run the numbers, model the levers

Funnel Math

Lead Velocity Calculator

Model the impact of response-time improvements on monthly closed transactions. Input your current latency, target latency, lead volume, and conversion baseline. Outputs the lift.

60 sec →
Unit Economics

Customer LTV Calculator

Compute lifetime value, LTV:CAC ratio, payback period, and a 24-month cohort projection. Real-estate preset uses GCI per side and repeat-and-referral expansion as the LTV driver.

90 sec →
Voice AI

Voice Agent ROI Calculator

The headline calculator behind Module 01. Drag the sliders to see the revenue an AI voice agent recovers from missed and after-hours calls. Twelve-month projection. Real-estate preset.

60 sec →

Want the brokerage stack running this quarter?

We white-glove install the 5-module engine for brokerages with a clear lead source already in place. Setup is 21 days on the standard track, 14 days on the priority track. Pricing scales with agent count and routing complexity.

See Pricing → Book a Walkthrough
N1 $444 / mo N2 $1,500 / mo N3 $4,000 / mo

Brokerage Operator FAQ

Will this integrate with our existing IDX and MLS feeds?
Yes. The engine ingests inbound leads from any IDX-fed source via webhook or polling. Most major IDX providers and MLS portal lead capture flows (RESO Web API, IDX Broker, iHomefinder, Showcase IDX, Zillow Premier, Realtor.com Connections Plus) are supported out of the box. For showing scheduling, Module 03 reads MLS showing instructions either via ShowingTime, Aligned Showings, BrokerBay, or a direct RESO Showing API where available.

For brokerages on a less-common MLS or a private feed, we ship a 48-hour custom connector build as part of standard onboarding. No additional cost on N2 or N3 tier.

Does the AI need a real estate license to do this?
The AI is configured as non-licensed support staff. It handles intake, scheduling, confirmation, and routing. It never gives price opinions, never advises on contracts, never represents the buyer or seller, never performs any act that requires a license under state real estate law. Anything license-bound is explicitly routed to a licensed agent on your team.

Every script ends with "I'll connect you with [Agent Name] who's licensed in [State] and can answer that for you." This posture is documented in the BAA and reviewed against the specific state real estate commission rules in every state we operate.

How do you handle Fair Housing compliance?
Every conversation flow is built with explicit Fair Housing guardrails. The AI never references protected characteristics under the Fair Housing Act (race, color, religion, national origin, sex, familial status, disability, plus state-level protected classes where applicable). The AI never discusses schools as a proxy for demographics (a documented steering risk). Neighborhood discussion is capped at geographic and functional descriptors only (ZIP, distance to landmark, lot size, walkability score from a neutral source).

Scripts are reviewed by a Fair Housing compliance attorney before deployment. We document the posture in the BAA. Any custom script changes require a re-review. The audit log preserves every AI utterance for review if a complaint is ever raised.

What about RESPA and lender introductions?
Module 02's preferred-lender intro flow triggers a RESPA-compliant Affiliated Business Arrangement disclosure (12 CFR 1024.15) automatically before any lender contact happens. The lead receives the ABA disclosure as a signed PDF via secure portal, must acknowledge, and the acknowledgment is logged in the CRM for auditability.

If your brokerage doesn't have an ABA with a lender, the AI simply routes lender questions to the licensed agent for a generic referral with no Section 8 RESPA exposure. We won't let the AI participate in an arrangement that creates kickback or referral fee risk under RESPA Section 8(a).

Does this work for teams inside a larger brokerage, or only standalone brokerages?
Both. The default deployment is at the brokerage level (single brand, single roster). For teams operating inside a larger brokerage under a DBA or team brand, we configure separate agent constellations, separate caller ID, separate scripts, and separate compliance posture. The team's branding and CRM are isolated from the broader brokerage.

For multi-team brokerages where each team wants its own AI agent setup, the N3 tier supports multi-team routing out of the box. Each team gets its own number, its own qualification script, its own scheduler config, and its own past-client nurture cadence, all under one brokerage roof.

How is SMS A2P 10DLC compliance handled?
Every SMS flow is A2P 10DLC compliant. Brand registration, campaign registration, and sender ID provisioning are done as part of standard onboarding. Every outbound SMS includes STOP and HELP keywords. Frequency disclosure is included on the opt-in (msg frequency varies, msg/data rates may apply). Phone numbers are never shared with third parties for marketing purposes.

Opt-in capture happens at the first lead intake point (web form, voice agent confirmation, or open-house sign-in). The exact opt-in language is in our SMS Privacy & Terms. Replies to any system message that aren't STOP, HELP, or workflow keywords route to a live person for handling.

Can the AI handle dual agency situations?
Where dual agency is legal in the brokerage's state and disclosed properly, the AI flags any intake conversation where the inquiring buyer is asking about a listing the same brokerage represents. The conversation never proceeds without explicit licensed-agent handoff and the required dual-agency disclosure language. In states where dual agency is prohibited or restricted (Florida transaction broker model, Colorado required disclosures, etc.), the AI is configured to route to a separate buyer's agent from a different team within the brokerage rather than the listing agent.

State-by-state dual agency rules are encoded in the routing logic and reviewed annually with the brokerage's designated managing broker.

What happens if the AI gets a question it can't answer?
By design, the AI is configured to admit it doesn't know and route to a human. Any question that touches price opinion, contract advice, legal interpretation, financing specifics beyond a simple referral, or anything the AI's training doesn't confidently cover, gets a consistent response: "Great question. That's one I want our licensed agent [Name] to answer for you. She'll call you back inside [time]. Anything else I can help with while we schedule that?"

Refusal is a feature, not a bug. Brokerages running this stack report higher trust scores from leads precisely because the AI doesn't try to bluff. Leads feel they've been heard, not handled.