How to Build an AI SDR That Actually Works | S1E2

“We had 30 people manually building prospect lists. Now Clay does it in two hours.” Bill Stathopoulos, CEO of SalesCaptain, isn’t talking about hype—he’s talking about GTM engineering that actually ships revenue. In this episode, Bill screen-shares the exact Clay table his team uses to find spa owners who don’t exist on LinkedIn, identify which booking software they’re already using, and automate what used to take SDRs 4 minutes per company. The kicker? This workflow processes 4,000 companies per week. No theory, just the actual production system.

Brought to you by Skylar, the AI sales coach - https://www.getskylar.com/
Bill’s LinkedIn: https://www.linkedin.com/in/bill-stathopoulos/
SalesCaptain: https://www.salescaptain.com/
Ivelin’s LinkedIn: https://www.linkedin.com/in/ivelin-kozarev/

[00:00] Cold Open: The $1M Outbound Campaign
How Bill went from creative agency side hustle to full-time GTM operations

[01:54] Why ZoomInfo Can’t Compete with Custom GTM Engineering
Static databases vs. AI-powered enrichment workflows that find what doesn’t exist in the “universe”

[03:15] The Real AI SDR Use Case
Forget AI-generated personalization—use AI to segment so precisely your messaging doesn’t need to be “personalized”

[05:02] Finding Business Owners Who Aren’t on LinkedIn
The booking software problem: How do you reach massage therapists and salon owners?

[08:57] Live Clay Table Walkthrough
Bill screen-shares the production workflow: Google Maps scraping, website crawling, tech stack detection, and owner extraction

[11:24] The Old Way: 4 Minutes Per Company
What SDRs used to do manually, now automated across 4,000 rows in 1-2 hours

[13:14] From 250 to 4,000 Weekly Touches
Why AI SDRs aren’t about replacing humans—they’re about 16× capacity multiplication

• GTM engineering beats static databases — ZoomInfo has companies; Clay builds custom enrichment engines that find SEC broker numbers, multi-location status, and owner emails scraped from “Contact Us” pages
• The best AI SDR work is invisible — Don’t over-personalize. Use AI to segment so well (eyewear brands see eyewear case studies) that your messaging is automatically relevant
• Build your own lead generation engine — Scrape Google Maps + TripAdvisor, use Claygent to extract owner info from websites, detect competitor tech stacks, normalize data—all in natural language prompts
• Scale = time compression, not replacement — What took an SDR 4 minutes per company now runs on 4,000 companies in 2 hours. That’s not hype, that’s math.
• Production workflows take 2 weeks to build — But once live, they run indefinitely. Bill’s team iterated criteria over time, then scaled to hundreds of thousands of companies.

Most “AI SDR” demos are vaporware. Bill’s team replaced 30 manual researchers with Clay workflows in 2024—not because AI writes better emails, but because AI eliminates the grunt work that makes segmentation impossible at scale. Their booking software client needed to reach spa and wellness owners (people who don’t use LinkedIn). The play: scrape local directories, visit websites to find owner names on contact pages, enrich with Instagram handles and competitor software, then trigger messaging based on tech stack. Result: 4,000 hyper-targeted touches per week vs. 250 from a traditional SDR. This is GTM engineering in production, not a proof of concept.

Episode Highlight: Bill actually opens his Clay table on screen and walks through each column—Google Maps scrape, AI agent prompt to find owners, tech stack detection, review count as proxy metrics. If you’ve been wondering what “GTM engineering” looks like in the wild, this is it.

Chapter TimestampsKey TakeawaysWhy This Matters