Speed-to-lead in seconds: 3× more conversations

A high-volume brokerage lost inquiries while agents were out showing properties. We built an outbound-and-inbound qualification agent that responds in seconds, asks the right questions, scores intent, and warm-routes hot leads straight to an available agent, with every detail captured in their CRM.

<5s
Lead response
3×
Qualified conversations
100%
Leads captured

What they came with

Portal enquiries arrived while the agents were out on viewings, and the callback came hours later. By then the buyer had already spoken to whoever rang first, and a lead the team had paid for was spent. Enquiries that landed in the evening or over a weekend waited until the next working morning, which is the same problem with a longer gap.

What the engagement covered

  • Instant response that catches intent at its peak
  • Conversational qualification scored against their criteria
  • Warm hand-off and automatic CRM capture

Technical detail

Webhook acknowledged, dial queued

A new lead in HubSpot fires a webhook that is acknowledged immediately, with the dial placed from a queued worker rather than inside the request. The telephony call never sits in the webhook response, so a slow provider cannot stall the acknowledgement and trigger a retry storm.

Qualify before touching the diary

Budget, area and timeline are captured as structured fields and checked against the booking rules before availability is queried. A viewing cannot be written to the diary from a conversation that has not cleared all three.

Held slots expire if the call drops

The agent takes a short hold on the slot and only confirms it once the caller agrees, and the hold expires by itself if the line drops mid-booking. Nobody has to clear phantom viewings out of the diary the next morning.

Bounded retries, then a human

No answer and voicemail route the lead into a fixed retry schedule and then to a person, rather than leaving the agent redialling. Every attempt, disposition and recording is logged against the contact, so the next caller can see what has already been said.

The stack

Telephony

Twilio Programmable VoiceTwilio Media Streams

AI

GPT-4oOpenAI Realtime API

Backend

Node.jsTypeScriptRedisPostgreSQL

Integrations

HubSpot CRMHubSpot WebhooksHubSpot Meetings
Practice
Lead Qualification
Sector
Real Estate
Shape
Client engagement
Stack
Anthropic, LiveKit, Redis
Next case studySupport chats resolved by AI and escalated with contextAn online outdoor apparel retailer was drowning in order status, returns and billing chats that queued for hours. We built an AI support assistant grounded in their Orders and Payments APIs that answers with live tracking, starts returns and emails labels on its own, and hands billing disputes to a human with the full transcript, order and payment history attached, so nobody repeats themselves.

Something like this to build?

Tell us what runs today and where it hurts. An engineer reads it and replies.

Start a projectAll case studies