Build the “Facebook Marketplace for AI agents” — broker deals between agents (and AI hiring humans), own the ranking, safety review, and distribution, and take a 10-30% cut of every match.
Dan puts it at the top of the A band: “Almost nobody is doing this at scale — there's no eBay for AI-to-AI.” Once a marketplace catches, it's fiercely defensible (Uber/Airbnb/Instacart), but the cold start is brutal and the take rate is good money, not insane money.
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
[Video] Dan: you go build the “Facebook Marketplace for AI agents” — you decide what has value, how it ranks, how safety gets enforced, and you market the marketplace itself. The Airbnb of AI agents.
A marketplace has no first customer, it has a first pair, and the only workable cold start here is to stop calling it a marketplace for a while and run it as a matchmaker.
The quick-reference card — acquisition channels, replication playbook, and risk map — is members-only.
Join the Unicorn Club — $5/moCancel anytime, in one clickPaste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself.
Data credibility: ⚠️ An opportunity direction, not a specific company; Dan's take rate is a hypothesis and the market hasn't formed yet. All revenue figures are founder-reported or estimated by the video creator — unaudited.
The first money in a marketplace rarely arrives as a take rate. It arrives as a brokered invoice: the operator sells an outcome to one buyer, hand-picks the agent that delivers it, pays that agent, and keeps the spread.
The turn in a marketplace is liquidity, and it is measurable rather than a feeling. It is the point where a posted job gets filled without the operator calling anyone, and where a buyer's second request arrives before the operator follows up.
The quiet stretch is the whole business here, and it runs longer than in any service model on Dan's list. He says as much when he ranks it: building this would “probably be harder to do that than start a venture studio.”
The rest of this breakdown — 7 more sections — is in the Unicorn Club, with every section of all 1,335 ideas: first customers, first dollar, the flywheel, and the silent stretch.
Join the Unicorn Club — $5/moCancel anytime, in one click