Every Monday I break down one AI system that runs inside my businesses. Real systems, real work. Here's #1.
The problem
Clickstack gets enquiries all day — from the website form, from WhatsApp, from Instagram DMs. Half are ready to talk budgets. Half need a case study, a call-back time, or just aren't a fit. My team was triaging all of it by hand, and the ready-to-buy leads waited behind the browsers.
The system now
- Every enquiry lands in one place. Form fills, WhatsApp messages and logged calls all arrive in a single queue.
- An AI agent reads it and tags it: ready-to-talk / needs-information / not-a-fit.
- Ready leads go straight to a human with a pre-drafted reply attached.
- Needs-information cases get an automatic message with the relevant case study or the roadmap.
- Nothing gets sent to a prospect without human approval. The agent prepares; people decide.
The result
My team starts every morning with a sorted queue. The fastest response usually wins the lead, and now the fast response goes to the right lead.
The principle
Agents should research, prepare, qualify and draft. Humans approve anything a client sees. That one rule is why this works and most "AI automation" doesn't.
Next Monday: the follow-up agent.