Case study / Property / AI Agents for Ops & Support

First response in under two minutes, and 63% more viewings booked.

Property enquiries from portals, the website, and WhatsApp were waiting hours for a reply. An AI assistant now responds instantly, qualifies budget and timing, books viewings into negotiators' diaries, and hands hot leads to the right person.

Industry: Property
Client: Independent estate agency
Scale: 4 branches · 19 negotiators · ~1,900 enquiries per month
Timeline: 4 weeks from discovery to all branches
Solution: AI Agents for Ops & Support

4.5 hrs → 2 min

first response time

71%

of enquiries handled without a negotiator

+63%

viewings booked per month

14 hrs

admin returned to each negotiator weekly

01 / The challenge

Where the time was going.

Enquiries arrived from Rightmove, Zoopla, the website, and a shared WhatsApp number. Negotiators picked them up between viewings and valuations, often that evening. By then many applicants had booked with a competitor.

01Average first response of 4.5 hours, and over 12 hours for weekend enquiries
02The same qualifying questions asked by hand hundreds of times a week
03Viewing requests bounced back and forth over email to find a slot
04CRM records incomplete because updates were typed up at the end of the day
05No way to tell which marketing channel produced applicants who actually offered

02 / The solution

How the workflow runs.

We put a single assistant behind every enquiry channel. It replies in the channel the applicant used, gathers the essentials, and books the viewing while interest is highest.

I
Respond instantly, in channel
Portal leads, web forms, and WhatsApp messages all receive a personalised reply within seconds, referencing the specific property.
II
Qualify conversationally
The assistant confirms budget, position (first-time buyer, chain, cash), timescale, and mortgage status in a natural back-and-forth rather than a form.
III
Offer real viewing slots
It reads the listing negotiator's calendar, respects travel time between properties, and offers slots that fit. Confirmations go to applicant and negotiator.
IV
Score and route
Each applicant is scored on fit and readiness. Hot leads are handed to a negotiator with a full summary; the rest enter a nurture sequence with similar properties.
V
Keep the CRM honest
Every interaction, answer, and booking is written to the CRM in real time with the originating channel attached.

Human-led by design

Guardrails that kept people in control

The assistant never negotiates price or discusses offers. Those conversations go to a person.
Vendors' private information and access arrangements are never shared until a viewing is confirmed by a negotiator.
Applicants can request a human at any time, and are always told they are talking to an assistant.
Branch managers review a weekly sample of conversations for tone and accuracy.

Stack

Rightmove & Zoopla lead feeds
WhatsApp Business API
Anthropic Claude
Reapit CRM
Google Calendar
Looker Studio

Timeline

Week 1
Discovery
Reviewed 600 recent enquiries, defined the qualification script with senior negotiators, and set scoring rules.
Weeks 2–3
Build
Connected channels, CRM, and calendars; ran the assistant in shadow mode drafting replies for negotiators to approve.
Week 3
Pilot
Live on one branch for web and WhatsApp enquiries, then portals.
Week 4
Rollout
All four branches live with weekly conversation reviews.

03 / The results

Before and after.

Measured over the 90 days after go-live against the 90 days before. Figures are rounded and client details anonymised.

Measure	Before	After
First response time	4.5 hours	Under 2 minutes
Enquiries handled end to end by the assistant	0%	71%
Viewings booked per month	410	668
Enquiry-to-viewing conversion	22%	35%
CRM records complete at first contact	~40%	98%