A multi-agent estate agency was finding listings through WhatsApp prospecting. It worked. Then the replies, the opt-outs and the admin started to pile up faster than anyone could handle them.
Every reply had to be read by hand. To find the few people actually thinking of selling, someone had to open and judge every single conversation.
One login, shared by everyone. Giving each agent their own licence was too expensive, so campaigns ran from one shared account. Everyone could see everyone's leads.
Messages went out from the wrong number. With several phones on one account, a campaign meant for one agent's number would send from another, and the database got harder to untangle after the fact.
Opt-outs lived in a spreadsheet. Each agent needed their own do-not-contact list, checked by hand before every upload. One miss means messaging someone who asked you to stop.
Two agents, same homeowner, same week. Nothing stopped overlapping lists from reaching the same person twice, which annoys prospects and starts fights over who owns the relationship.
Leads were retyped into the CRM. The agency's CRM has no API, so every good lead was copied across manually.
messages to read from a single campaign: 216 prospects, three-message sequence, all reviewed by hand to find the hot leads.
The conversation started with licence costs and a clunky interface. In the workshop it became clear those were symptoms. The business depends on one person setting up campaigns and doing the admin around them. In a busy week, that admin simply doesn't happen, and a week of prospecting is lost.
We also learned what had to stay human. In residential property, the relationship starts the moment a homeowner replies, and it has to come from the agent's own number in the agent's own voice.
“The relationship-building starts the moment they reply. That's not something you hand to an AI.”
So the brief became clear: let software do the sorting, checking and admin, and keep every conversation personal. Doing that also meant one platform licence could serve the whole team, instead of one per agent.
The agency's existing outreach platform protects numbers from being blocked, which agents value most. We kept it as the engine and built a branded portal in front of it. One licence, one WhatsApp number per agent, and a separate login for everyone.
A guided flow replaces the old interface. Pick a goal, upload the list, and AI drafts the three-message sequence with tips beside each message. Every campaign sends from the right agent's number.
Goal → Upload leads → Write messages → Schedule → Launch
Fixes: wrong-number sends, clunky set-up
All replies land in one inbox. AI scores each one from 0 to 100 with a one-line reason, flags what's waiting on the agent, and catches opt-outs automatically. Agents still reply from their own phone.
Fixes: reading every reply by hand
Anything scoring 70 or more moves to Hot Leads on its own. One click exports the new ones to Excel, so the agency's existing CRM gets clean data without retyping.
Fixes: retyping leads
Uploads are checked against one company-wide Do Not Contact list. If another agent messaged the same person recently, they're held on the Shotgun list until the cool-off passes, and an admin resolves any clash.
Fixes: spreadsheet opt-outs, agents competing for the same lead
Agents see their own work. Branch owners compare branches and agents side by side in plain English. Admins see the whole company, plus an audit log of who launched, exported or overrode what.
Fixes: one shared login, no visibility
Next on the roadmap: a follow-up diary, so a “maybe next year” reply becomes a dated reminder instead of a forgotten lead.
Screens from the working prototype, shown with sample data.
Most teams already know where the hours go. We help you find where the value is, then show it working before you commit to a build.
We map the manual work, the tools and the real cost.
You click through a working version built around your process.
Clear features, priorities and a fixed estimate.
We develop, launch and support the system.