Automating Lead Follow-Up & Routing with AI

Client context

A US business running its sales pipeline on GoHighLevel (GHL) generated a steady stream of inbound leads: form fills, ad responses, inquiries. Like most teams at that stage, follow-up depended entirely on sales reps noticing new activity and remembering to act on it.

I was brought in as the automation engineer on the engagement to answer one question: could follow-up happen instantly, on every lead, without hiring more people?

The problem

Speed-to-lead was the killer. Leads that didn't get a response within hours went cold, and reps juggling calls simply couldn't keep up. The team's existing drip campaigns didn't help much. They were static: the same generic sequence regardless of what the lead had actually done.

A lead who had opened three emails and clicked a pricing link got the same message as one who had never engaged. There was no routing logic reacting to behaviour, no personalization beyond a first-name merge tag, and no coverage at all outside working hours.

The automation I built

I designed and built an AI Follow-Up Agent orchestrated in n8n, sitting on top of the client's existing GoHighLevel account (no CRM migration required). The key workflow logic:

• Behaviour triggers: n8n listens to lead interaction events from GHL, including new lead created, message opened, link clicked, call missed, and form submitted, and routes each event into the appropriate branch of the sequence.
• Contextual message generation: instead of static templates, OpenAI generates the follow-up content using the lead's actual context (source, stage, and interaction history), so every touch reads like it was written for that person.
• AI voice calls: for high-intent moments, the agent places outbound calls through Twilio with natural-sounding speech generated by ElevenLabs, delivering a personalized voice message rather than a robocall script.
• Multi-channel sequencing: calls, SMS, and email are coordinated in one state machine: the agent knows what was already sent on which channel and escalates or backs off accordingly.
• Behaviour-based branching: engagement changes the path. A lead who replies gets routed to a human immediately; a quiet lead gets a different cadence than an engaged one. This is routing driven by behaviour, not a fixed timer.

Quantified outcome

The agent turned follow-up from a rep-dependent chore into infrastructure that runs around the clock.

• Every lead now receives timely, personalized follow-up across three channels (call, SMS, and email) with zero manual sequence management.
• Response coverage went from business-hours-if-someone-is-free to 24/7: a lead arriving at 2 a.m. gets the same instant treatment as one arriving at 2 p.m.
• Sales reps were freed from mechanical touches entirely, engaging only when a lead responds, which is the moment their time actually matters.
• Follow-up quality stopped depending on rep workload: the busiest week produces the same consistency as the quietest.

Stack