Each agent was spending 3+ hours a day on WhatsApp copy-paste. We gave them that time back.

Agents doing the work a system should have been doing.

Priyanka, Burj View's founder, tracked her agents' WhatsApp time for one week. The result was worse than she expected: each agent was spending an average of three hours and twenty minutes every day on tasks that were, in her words, "identical every single time." Greeting a new Bayut lead. Copying their details from the portal into the CRM. Sending a follow-up message at 24 hours. Sending another at 48 hours. Updating the CRM stage after each touch.

None of this required an agent's judgment or expertise. But it was consuming the hours that should have gone to viewings, negotiations, and actually closing deals. The eight agents at Burj View were collectively spending over 130 hours a week on structured WhatsApp admin.

There was also a consistency problem. Whether a follow-up went out at 24 hours or 72 hours depended entirely on how busy the agent was that day. Some leads got two follow-ups. Some got none. The outcome was inconsistent because the process was human-dependent.

How we ran the pilot.

Days one and two we mapped every repetitive WhatsApp task in detail: what the agent typed, when they typed it, which CRM field they updated after. We ranked each task by time cost and how easily it could be automated without losing the personal quality of the conversation.

Days three through seven we built the core system: WhatsApp Business API connected to a shared inbox, n8n workflows pulling Bayut and Property Finder leads automatically, Anthropic-powered AI handling first contact and qualification, and the 24h and 48h follow-up sequences written and scheduled. Every sequence was personalised using the lead's specific property interest and budget range extracted from the conversation.

Days eight through ten we ran live Bayut leads through the system alongside agents. They watched the AI send the messages they would have sent themselves. Several asked to see the follow-up templates and suggested small tweaks in tone. We updated them on day nine.

What we shipped.

AI handles first contact on every Bayut and Property Finder enquiry. The Anthropic-powered layer sends the opening message, asks the qualifying questions, and extracts budget, timeline, and area preference before the agent is even notified. Agents step in once the lead is warm.

24-hour and 48-hour follow-ups run without anyone remembering. Every unbooked lead gets a personalised follow-up at exactly 24 hours and again at 48 hours. The messages reference the specific listing they enquired about and include a Cal.com booking link. Agents do not touch these.

Zoho CRM updated after every message automatically. Lead source, qualification score, conversation stage, and follow-up status are written to the CRM by the system. No manual entry, no forgotten updates.

Agents only handle conversations that need a human. Anything requiring negotiation, property knowledge, or buyer judgment gets flagged to the assigned agent with a summary. The routine 80% never reaches their queue.