Cold Leads Pipeline Automation

Using StackAI, I built an automated cold deal re-engagement pipeline to streamline one of the more common revenue leaks in early-stage startups, which are deals that go cold because follow-ups fall through the cracks.

During an interview process with an early-stage startup, the CEO told me their biggest needs was operational efficiency, including their GTM processes. As they scaled their commercial motion, manual processes were going to become impossible to sustain. That conversation stuck with me.

This conversation happened to come one week after StackAI came to MIT Sloan to demo their product and saw an opportunity to test a new tool against a real problem.

• Cold Deal Detection: A 9AM job queries a Notion CRM for deals where Last Contact Date exceeds 28 days, flagging cold leads automatically.
• AI Email Drafting: Deal data is passed to Claude, which drafts a personalized follow-up for each lead and returns structured JSON that maps directly to the CRM schema.
• Human-in-the-Loop Approval: Drafted emails are written to a second Notion database and the GTM rep is notified via Gmail. Nothing moves without checkbox approval, which is the only manual step in the pipeline.
• Automated Send & CRM Update: A 12PM job reads approved drafts, sends HTML-formatted emails from the CEO's Gmail, updates the CRM, and archives sent drafts, closing the loop completely.
• Tech: StackAI · Notion API · Anthropic Claude API · Gmail API · Python

I presented this automation in the 3rd round interview and received an offer. More importantly, I came away with a repeatable architecture for making outbound communications more efficient using AI that can be applied to any GTM, patient outreach, or operational follow-up workflow.