An AI agent that qualifies leads the moment they land
A conversational agent that engages new inquiries instantly, qualifies budget and intent, books viewings, and hands warm leads to agents with full context.
The brief, in specifics
Residential Brokerage
8 weeks
Fixed price, phased
Client name withheld under NDA. Engagement details are shown to the extent our agreement permits.
What we were brought in to do
Weekend and after-hours leads were lost because no one replied fast enough. We built an AI agent that responds in seconds, qualifies the lead, and books the viewing.
A twelve-agent brokerage where inquiries arrived at all hours and were answered during office ones. The owner had measured it before calling us: the median first reply was four hours, and on weekends it was nineteen. Buyers in this market contact three or four listings at once, so a reply that arrives fourth arrives to someone who has already booked a viewing elsewhere.
AI & Automation
Where the old way broke
Inquiries arrived around the clock but agents replied in hours. By then, prospective buyers had moved on to competing listings.
The loss was concentrated and measurable. Of the inquiries that never converted, sixty percent had received a reply more than two hours after arriving, and in the follow-up calls the brokerage ran, the most common reason given was that they had already seen something. It was never a qualification problem. It was a latency problem being treated as one.
We deployed a qualification agent across web chat and SMS that engages instantly, asks the right qualifying questions, syncs to the CRM, and routes hot leads to agents with full context.
Instant Conversational Intake
Responds in under 30 seconds via SMS and web chat to qualify buyer budget, timeline, and location preference.
The agent answers within thirty seconds on whichever channel the lead arrived through (Twilio for SMS, a widget for web) and works from a fixed qualification script instead of open-ended chat, so budget, timeline and area come back in a shape the CRM can store as fields. Every reply is grounded in the live listing set, so it never offers a property that has already gone under offer.
- Under-30-second first reply on SMS and web chat
- Fixed qualification script that returns structured fields
- Answers grounded in live inventory
The buyer's side of a Friday-night text: a reply fourteen seconds later confirming the house is still available, then timeline before budget, each answer tagged with the field it fills.
Smart Viewing Scheduler
Saturday by negotiator: fourteen viewings, twelve booked by the agent in conversation, beside the Orchard Rise booking, where hold, event, confirmation and two reminders commit as one transaction.
Direct calendar integration allowing buyers to book property viewings instantly with automated SMS reminders.
Availability comes from each agent's own calendar through Cal.com, so a slot offered in conversation is a slot that genuinely exists. Booking writes the event, sends the confirmation and schedules two SMS reminders as one transaction; a partial failure releases the hold instead of leaving a phantom viewing on someone's calendar. Reschedules run down the same path, so there's no second, divergent flow.
- Live agent calendars via Cal.com, so no phantom slots
- Book, confirm and schedule reminders in one transaction
- Reschedules reuse the booking path
CRM Synchronization
Syncs buyer intent, budget, and conversation logs straight into the CRM for agent follow-up.
Every conversation lands in HubSpot as a structured record: parsed budget, timeline and area written to the contact, the full transcript on the timeline beneath it. Deduplication runs on phone and email before the write, so the same buyer texting and then filling the web form doesn't become two leads. Hot leads trigger owner assignment with the qualifying answers already inside the notification.
- Parsed fields on the contact, transcript on the timeline
- Phone and email dedup before write
- Owner assignment carrying the qualifying answers
One contact record: four of the six answers written as fields and two still open, the web chat merged with an earlier text from the same number, and eight messages ending with the lead passed to its negotiator.
What we built together
- 01
Modeled the questions top agents ask to qualify a buyer
Timeline comes before budget in the script, an order taken straight from the brokerage's two best closers.
- 02
Wired the agent to the CRM for two-way sync and routing
Every conversation writes parsed fields to the contact and the full transcript to the timeline, with deduplication on phone and email before the write.
- 03
Added calendar booking with automatic reminders
Availability comes from each agent's own calendar, so a slot offered in conversation is one that exists, and booking and reminders happen in a single transaction.
- 04
Set escalation rules so complex cases reach a human fast
Escalation rules were tuned weekly against real transcripts, and the brokerage decided what should have reached a human. The model didn't guess.
Operational results after launch
< 30s
Lead response time
+46%
Qualified leads
+29%
Viewings booked
Response time is measured from inquiry receipt to first outbound message, across every channel, including the ones the agent doesn't handle. Qualified leads and viewings compare the three months after launch with the three before, using the brokerage's own definition of qualified, never the agent's own scoring.
Client name withheld under NDA. Figures are approximate, drawn from the engagement’s own reporting.
Phase by phase
Phase 1: Agent Logic Design
Conversational Qualification Specs
Modeled top agent intake scripts to craft an empathetic conversational flow qualifying buyer budget, timeline, and property criteria.
- Qualification Script Specs
- Fair Housing Compliance Rules
- Lead Scoring Matrix
Phase 2: CRM & Calendar Engineering
Two-Way Integration Pipeline
Built two-way API connectors linking Twilio SMS and web chat directly into the CRM and live agent calendars.
- CRM Connector
- Cal.com Scheduling Pipeline
- Lead Context Summarizer
Phase 3: Multi-Channel Deployment
Web Chat & SMS Voice Agent Launch
Deployed the AI agent across the website chat, SMS shortcode, and off-hours phone redirects with instant response handling.
- Web Chat Widget
- SMS Intake Handler
- Call Redirect Bridge
Phase 4: Handoff & Protocol Tuning
Agent Handoff & Performance Review
Tuned human-in-the-loop escalation triggers so warm leads immediately alert the assigned agent via SMS and CRM push notifications.
- Escalation Protocol
- Sales Staff Training
- Lead Conversion Analytics
Ten leads against the six questions that decide a viewing, which questions go unanswered most, and that nothing was booked on fewer than five.
About our collaboration
- 01
A cross-functional team of 4 worked on a fixed price, phased basis over 8 weeks, covering AI agent design, CRM integration, Handoff workflows. We ran a weekly demo and kept a shared board open to check at any time. An in-house team took over day-to-day operation before the engagement ended, with handover built into the last phase.
The qualification script came from recording the brokerage's two strongest agents, which is why it asks about timeline before budget: both of them did, and both said the order mattered. Escalation rules were tuned weekly for the first six weeks against real transcripts, with the brokerage's own agents deciding what should have reached a human.
What we'd carry into the next one
Sub-30 second response times captured 46% more qualified leads than manual agent outreach.
Speed was the whole mechanism. The same questions asked four hours later converted at the old rate, which is how we know the problem was latency.
Automated calendar booking increased completed property viewings by 29%.
Viewings rose because booking happened inside the conversation. The previous flow ended with a promise to call back, which is where most of them stopped.
2-way CRM synchronization eliminated duplicate lead data entry for sales agents.
Structured fields, not a transcript summary, removed the retyping. They're also what made the brokerage trust the lead scores enough to work them.
Reply in seconds, qualify, then book the viewing or hand it to a person
A Friday-night text about a house with a board outside, from the first reply to a Saturday viewing. Pick a step to see what the agent did and why. Use the arrow keys once a tab is focused.
From an inquiry landing to a viewing on the calendar
The questions were never the problem; the four-hour wait was. Each stage keeps the buyer in one conversation until a viewing exists or a person has the lead.
- 01 · SourceWeb chat, SMS, off-hours call redirectEvery channel an inquiry arrives on is handled at any hour, including the ones outside office hours.
- 02 · IntakeChannel handler, instant replyFirst reply inside 30 seconds, on the channel the enquiry came in on.
- 03 · EngineFixed qualification scriptTimeline before budget, answers parsed into fields, replies grounded in live listings so nothing under offer is offered.
- 04 · StateCRM contact + timelineDedup on phone and email before the write; parsed fields on the contact, full transcript on the timeline.
- 05 · DeliveryNegotiator calendar + handoffBooking, confirmation and 2 SMS reminders commit as one transaction. Hot leads alert the owner with the answers attached.
Real slots, live listings, one record per buyer
How many buyers booked elsewhere while your weekend enquiries waited? Scope your build in 3 minutes.
Scope your buildNearby engagements
Web PlatformsA property portal built around how buyers browse
A property listing and search portal that helps buyers and renters find the right home fast, with a listing manager built for agents behind the scenes.
Residential Brokerage · 14 weeks
AI & AutomationAI phone ordering that fills every table's takeout slot
A voice agent that answers every takeout call, takes orders straight into the POS, and upsells like a trained host, even at the dinner rush.
Restaurants & Hospitality · 9 weeks
AI & AutomationA private legal assistant grounded in verified precedents
A private knowledge assistant that searches internal case files and precedents, providing cited answers legal teams can verify in seconds.
Legal & Law Firms · 14 weeks
Let's talk
Running a large platform, shaping a first MVP, or getting a product ready for a funding round? Tell us where you are. We'll shape the process around it, and stay with you after launch.
