Most outbound lists get built the same way.
You open Apollo or Clay. You pick the obvious industry, set a headcount range and guess five job titles. Then you export a thousand contacts and stop there, because every credit costs money.
The trouble is that everyone selling into that market runs the exact same search. Same filters, same titles, same list. Your prospects are getting ten versions of your email every week.
So in my latest video I went the other way. Go wide first, then let AI do the narrowing.
I picked recruitment agencies as the test market and ran the whole thing from Claude Code.
Claude searched 400+ industry tags and surfaced categories I'd never have thought to click on, like HR services, executive search and temp help. Firms that do recruitment but tag themselves as something else. That took the market to 448,569 companies.
Then it pulled every decision maker at C-level, VP and director, instead of me guessing titles. The right person at a 10 person agency is rarely the same as at a 50 person one, and you can't know that up front. 85,965 contacts, which Claude then filtered down to the genuine ICP fits.
From there it built three segmented campaigns in Instantly, verified the leads and wrote and loaded the copy.
An empty terminal to launch-ready campaigns, and I never touched a spreadsheet.
All the prompts are in a free Notion doc linked in the description, so you can swap in your own ICP and run the same thing this week.
Tom