We found 5,581 leads with a simple prompt

****

I want to tell you a story and then point you to a free resource.

One of our clients sells medical supplies to people who can't get to a hospital fast. We started digging through their customer list to figure out who their best buyers actually were.

Turns out it was fishing boat owners.

Before we asked a single question, nobody on their team knew that.

Fishing vessels are brutally hard to find. Most databases don't track them well, and the ones that do charge a fortune for access. We asked AI to do it.

The prompt we used wasn't fancy. Five or six lines, nothing clever. But it was specific: find fishing vessels, in these regions, owned by people who own more than one boat, and give me a way to contact them.

Perplexity Computer ran with that. Six steps just to confirm it understood the ask correctly. Then 57 more steps autonomously digging through public records, registries, and databases.

What came back: 5,581 leads.

Not just names, either. Where they're located, who owns five or more boats, who owns four, three, two, one. Email addresses. Company names.

Even where our client's existing customer base overlapped with these new prospects, so we'd know where a warm intro might already exist.

It even wrote a first draft outreach email for each segment.

Then it gave us an assumptions tab. Basically a note that said: here's what I guessed to build this list, and if any of it's wrong, change it and rerun the whole thing.

AI isn't a one-shot vending machine. You check its assumptions, correct the ones that are off, and run it again.

None of this works if your ask is vague. If we'd typed "find me leads in the maritime industry," we'd have gotten a wall of nothing useful.

The fishing vessel prompt worked because it was narrow enough that AI could actually go hunt for something real.

That's the whole game with AI lead gen right now.

Most people go too broad, then wonder why the results are garbage.

I walk through the exact prompt, the full report, and how we built the whole thing in this 20 minute podcast episode I appeared on recently with Andrew Warner:

​[Learn how to do lead gen with AI]​

Talk soon,

Matt

P.S. One thing I don't want to bury here: AI didn't pick fishing boats out of thin air.

We found that market by going through the customer list first, and fishing vessel owners were already their best customers. High close rates, strong margins, and barely anyone on their team was actually chasing them.

That homework is what made the prompt sharp enough to produce a list this good.

Skip that strategy step, and you're asking AI to guess at a market.

Do it first, and you're asking AI to go deeper into one you already know works.