AI SMS lead vetting with scoring and call rounds

Deliverables

• Main inbound SMS vetting workflow with AI intent classifier
• Conversation AI qualification bot with adaptive questions
• Code-based lead scoring with Immediate, Hot, Warm and Cold buckets
• Lead vetting pipeline with per-bucket call round stages
• Round advancement and post-round SMS workflows
• Re-engagement waves and Slack alert workflow

The challenge

SMS campaigns to property owners produce a steady stream of replies, and most of them are not worth a call. Some are price quotes, some are wrong numbers or stop requests, many are a simple yes. Sorting them by hand is slow, sellers drop off halfway through a text conversation, and without a clear order the acquisitions team calls weak leads while motivated sellers wait.

Our approach

A GoHighLevel workflow starts when a contact replies by SMS. An AI agent step classifies the message as not interested or DNC, wrong number, referral, who is this, price quoted, interested or other, and each branch tags the contact and either ends or continues. Interested and price-quoted leads are handed to a Conversation AI bot running OpenAI GPT-4.1, which confirms ownership, timeline and asking price but skips anything the seller already said. In the background a webhook sends the address to Zapier for a PropStream property lookup. If the seller goes quiet, the workflow waits, sends one nudge, then carries on with the data it has. A custom code step combines price alignment, timeline and purchase date into a total score and returns a bucket of Immediate, Hot, Warm or Cold, plus a flag when value estimates disagree. Each bucket creates an opportunity in its own Round 1 stage of the lead vetting pipeline, and Immediate leads trigger a Slack alert.