Mailing Brass Fittings to Plumbers: The Dream 100 List’s Dumb Method That AI Can’t Replace

The Dream 100 list isn’t new. Chet Holmes used it to add 28 clients in a single month. A tweet from The Startup Ideas Podcast lays out a complete process.

The approach is specific: pick one trade in one city—say, plumbers within 25 miles of Dallas. List the 100 companies you’d most like to work with. Have an AI agent read reviews, scrape websites, and find the owners. Then go to Reddit to see how local customers talk about service providers and learn their language.

Then build a brand that belongs only to this market. The example in the tweet is called Leak Proof, with the tagline “We fix the leaks in your marketing; you stick to fixing pipes.” These owners may not respond to brand language like Stripe’s or Apple’s.

Next, help them for free every week. Run Meta ads for plumbers in this area, sending traffic to a free leak-inspection landing page. Then send physical mail to all 100 companies: put a brass fitting in the box, include a letter, link to the page you rebuilt for them, and invite them to a breakfast or plumbing/HVAC meetup you host.

The tweet says AI removes the hard parts. Record a short video ad and let Claude edit it; Claude Code can build campaigns through Meta’s MCP; you don’t even need an existing audience. Pick the trade, let the agent do research, then mail the boxes.

Someone in the replies points out a detail: the brass fitting is brilliant. Plumbers get 40 cold emails a week, and nobody sends them a part from their own trade.

Another reply comes from a construction company owner. He says the fastest way onto his list is to know what a change order is before you pitch. Most AI pitches can’t tell a pay app from a punch list.

Those two replies say more than any tool list. AI lowers the cost of research, pages, and ad creative, but it doesn’t solve industry language or persistent follow-up for you. The point of Dream 100 isn’t to pull a list and blast it. It requires focusing on a narrow market, showing up repeatedly, doing small favors, and waiting until they’re willing to talk.

In the video screenshot, Chet Holmes’s case is broken into six months: no sales for the first five months, then in the sixth month 28 people buy the entire product line at once. Yellow text at the bottom says most people quit before month five.

The really hard part of this method is whether you send the next box when nobody has bought in month five. For AI practitioners with a development background, agents can find leads, write first drafts, and build pages; they can’t understand change orders for you, and they can’t get you through the silent period. For ordinary readers, this is the reality of boring businesses: customers don’t care what model you use; they only care whether you understand their work.

AI can generate 100 emails that look personalized. The owner may just want to know whether you can tell a pay app from a punch list.