AI Sales Pipeline Automation

What it does

This skill runs cold outbound as a single automation chain instead of one-off scripts. The flow has six stages: (1) RB2B visitor ingest — resolves anonymous web traffic to a person/company and scores intent from the visited URLs via PAGE_INTENT_SCORES (e.g. /pricing + /demo = high intent). (2) 5-layer suppression — blocks outreach to the wrong person through existing-customer, active-sequence, opt-out, competitor and free-email filters, logging a rationale for each suppression. (3) Campaign routing — enrolls the surviving lead into the correct Instantly campaign by segment, including agency detection. (4) Dead-deal resurrection — deal_resurrector uses a 3-layer score (time decay + POC expansion + champion tracking) to turn HubSpot closed-lost deals back into reopenable pipeline and drafts revival emails. (5) Trigger prospecting — monitors buying signals (new hires, funding, agency searches) via Brave Search and scores them. (6) Self-learning ICP — icp_learning_analyzer mines approve/reject decisions for patterns and recommends filter changes, sharpening targeting over time.

The framework operationalizes "trigger on buying signals, not calendar reminders": signal-driven prospects reply far above average, and closed-lost prospects who already evaluated you convert roughly 5x better than cold outbound.

When to use

• Turning RB2B website visitors into automated outbound.
• Running mandatory suppression checks before cold email.
• Systematically reviving dormant closed-lost deals from HubSpot.
• Finding companies showing buying signals (new hires, funding, postings).
• Improving ICP targeting from approve/reject patterns.

Method / frameworks

Intent scoring is a URL-weighted intent model; suppression is a 5-layer guardrail sequence; resurrection scoring combines time-decay + opportunity-expansion + champion-tracking; trigger prospecting rests on a buying-signal taxonomy; ICP learning is a closed-loop optimization deriving filters from approve/reject feedback. Because a large share of closed-lost is driven by indecision (no-decision) rather than a competitor win, the resurrection layer is especially high-yield.