The problem: a beautiful site that didn't sell

Luxury real estate has a specific failure mode. The website is gorgeous — cinematic photography, virtual tours, impeccable typography. It's also silent. Visitors arrive from a listing, browse for a few minutes, and leave. No form fill, no phone call, no trace. The site is a brochure, not a business.

For Sotheby's International Realty, the gap showed up in three concrete problems:

  • High bounce, near-zero conversion. Qualified buyers hesitated to make first contact — a form fill felt like inviting a hard sell.
  • Missed after-hours demand. A meaningful share of high-net-worth buyers browse outside brokerage hours, when no one is available to respond.
  • A qualification bottleneck. Agents were spending their first call on the basics — budget, timeline, must-haves — before any real value could be discussed.

The mandate was narrow and measurable: turn the website from a cost center into a demand-capture asset. The tool was a conversational AI peer, deployed on HumaticAI's Planet 9 runtime.

What "qualified" means before you build anything

Most chatbot projects fail for a boring reason: nobody defines what a good lead is before the first line of dialogue is written.

The Sotheby's deployment started with that definition. A qualified lead had a verified budget range, an explicit timeline (usually within 90 days), a stated property type, and a preferred geography. Anything less was a "curious browser" — worth engaging, not worth a human's calendar.

This single decision shaped everything downstream: the questions the peer asked, the order it asked them, and the point at which it handed the conversation to a human.

The triage logic, step by step

The peer's job was not to answer "what's the price?" and wait. It was to run a structured, low-friction qualification without making the buyer feel interviewed.

  1. Intent detection. The first message tells the peer whether the person is browsing, selling, or actively buying. Sellers get routed differently from buyers.
  2. Structured discovery. The peer works through budget, timeline, and must-haves in conversational order, skipping questions the buyer already answered.
  3. Scoring and prioritization. Each conversation produces a lead score, so the sales floor sees the hottest inquiries first — not the ones that arrived most recently.
  4. Context-aware curation. Using live property data via HumaticAI Rover, the peer surfaces matching listings, virtual tours, and neighborhood context mid-conversation — a personalized showing before any human is involved.
  5. Warm handoff. When a lead crosses the qualification threshold, the peer packages the full interaction into a dossier and routes it to the right agent through the CRM MCP connector.

The point of this design: the human agent never starts from zero. They inherit a warm, documented, pre-qualified lead.

The 90-day timeline

  • Weeks 1–2 — grounding. The peer is trained on past client interactions, MLS data, and neighborhood specifics. This is where the knowledge base gets built, not where the lead flow starts.
  • Weeks 3–6 — soft launch. The peer goes live with monitoring. Responses and qualification logic get refined against real conversations. It learns to tell a "curious browser" from a "serious buyer."
  • Weeks 7–12 — full throttle. The peer handles more than 85% of initial website inquiries, around the clock. Human agents shift from first-contact triage to relationships, negotiation, and closing.

The results

The headline number is 200+ qualified leads in the first 90 days. But the more durable results are structural:

  • 24/7 capture. The peer erased time zones, engaging international buyers the moment they expressed interest.
  • Faster first conversations. The handoff dossier meant agents opened every call already several steps into the relationship, armed with the buyer's budget, timeline, and preferences.
  • Cleaner pipeline. Every interaction landed in the CRM through the MCP connector — no more guessing which leads were hot, no more orphaned inquiries.

On adoption, Serge Seregin, Member of the Board and CTO at Sotheby's International Realty, frames the shift plainly: the system took hold because it "did the work nobody wanted" — the data entry, the overnight replies, the dead-end qualification calls — and the sales floor noticed.

For the operational detail behind that adoption story, see Inside a Luxury Brokerage: What Happens When You Hire an AI Digital Employee.

Why this worked

Three things made this deployment work.

  1. The qualification criteria were defined up front, so the peer had a clear, measurable job from day one.
  2. The handoff was treated as the product, not an afterthought. The peer's purpose was to make the human more effective, not to hold a conversation.
  3. The rollout was opt-in and proof-driven, not mandated. Early adopters became the evidence that won over the rest of the floor.

Limitations and when this fails

The honest limitations are worth naming.

  • It multiplies a system that half-works. If the brokerage has no CRM discipline, the peer inherits the mess. It is a multiplier, not a cure for chaos.
  • It will not close a deal on its own. The peer qualifies and hands off a warm lead; a multi-million-dollar listing still closes through human judgment and rapport.
  • Brand voice needs work. A luxury brokerage cannot sound like a utility-company chatbot. The first week of any deployment involves correcting tone.

If a team has three agents who all know their clients by name, the overhead of a digital peer will not pay for itself. The economics work when there is volume: inbound inquiries, a large database, or geographic spread that makes manual follow-up impossible.

The framework to copy

The Sotheby's playbook reduces to four decisions:

  1. Define "qualified" first. Budget, timeline, authority — write it down before you build. The qualification logic follows from that definition.
  2. Design for the handoff, not just the chat. The AI's ultimate job is to make the human colleague more effective.
  3. Measure beyond lead volume. Track lead-to-meeting conversion, agent ramp-time reduction, and sales-cycle compression — not just "how many conversations."
  4. Launch where intent is highest. A contact page or a high-traffic listing page is a better first deployment than a full-site rollout.

For a broader look at how this fits a wider real-estate AI strategy, see AI Lead Qualification in Real Estate: Why Broken Triage Costs Brokerages Millions and AI SDR for Real Estate: How to Automate Lead Qualification.

The bottom line

A static website in luxury real estate is a strategic vulnerability. Sotheby's Realty closed the gap by treating the site as an always-on demand-capture engine, with an AI peer handling triage, qualification, and handoff around the clock. The 200+ qualified leads in 90 days are the headline; the structural shift — human agents doing the work they're paid for, instead of first-contact triage — is the point.

The question for brokerage leaders is no longer whether to make this shift, but how quickly.

See Planet 9 pricing. Digital Peers start free; Plus is $24/mo. Compare plans or Start free.