Prospecting on the public web is real research, and it works. But it's hunting with a flashlight. The only companies you can see are the ones making noise: press releases, job posts, business journal write-ups. The quiet ones are invisible from where you're standing.
Apollo can see them. The problem is Apollo's interface, which is a wall of filters that will make you close the tab. I open it on screen so you can see exactly what you're not going to learn, then close it and never touch it again.
Instead, one sentence to Claude: read my Who We Sell To page, translate it into an Apollo people search, and show me the translation in plain English before you run anything.
That's the whole episode, and it's the centerpiece of this track. The profile you wrote earlier stops being a document and becomes a live query against a couple hundred million people.
Watch how the conversation goes. Claude reads the profile out of Notion, comes back with its interpretation, and flags what Apollo can't do (there's no radius search, so "within 100 miles" has to become something else). Then it asks me two questions: owner-operator or professional CEO, and what the employee floor is for midsize. First run comes back with 3,528 matches and a list of suggested cuts.
Then the refinement loop, and here's the part that matters: searching is free. Only enrichment costs credits. So you can argue with the list as long as you want. Drop a city. Add an HQ filter. Companies between 50 and 500 people. Cut healthcare because those slots are full. Cut nonprofits and public institutions. 3,528 became 236, then 188, then 164, and none of it cost a thing.
Only when the list is right do you spend. Take the ten best fits, enrich them, write each into Targets as Waiting with name, company, title, why they match my profile, and source set to Apollo. Verified emails on their pages, not in the chat. Tell me the cost first and wait for my approval. Ten lead credits, approved, done. Ten strangers who should be your customers, sitting in the same queue as everything else.
Two honest notes. The demo data is contrived and I'm clicking fast, so your numbers will look nothing like mine. And the machine only pays off if the email actually gets sent, which is still your job.
Your assignment: get your market's real size, refine for free until the list is right, then spend ten of your 75 free credits and run one of those ten through the briefing and drafting machine you already built.