Founder note

AI-native services: pick three, prove willingness to pay

Opportunity maps list dozens of AI-native services. Pick three you can sell this quarter, and get proof that someone will pay before you build the rest.

Greg Isenberg recently argued that AI-native services are a very large opportunity: work that used to be sold as hours can now be delivered by agents with a person supervising. Maps like his are useful for seeing the space. They do not tell you what to sell on Monday. The hard part is choosing a few services you can deliver and charge for now.

The three services to start with

  1. Agent spend control for client work. Set budgets, caps, and usage receipts so a client never gets a surprise bill from agent usage.
  2. A workflow agent for one painful job in one industry. One job, one owner, and one handoff you can measure. Not a general assistant.
  3. Human review of agent output. A person checks the agent’s work at a defined point, so mistakes are caught before the client sees them.

Each one solves a problem buyers already feel, which makes the willingness-to-pay test fast.

How to use the list

Why it matters

Building a category map feels productive, but it doesn’t close a sale. Proof that a named buyer will pay is what separates a service from a demo.

Humoud’s take

If you can’t say who pays and what proof you’ll have within 14 days, the service is still a demo. Get the proof first, then scale.

Do this week

  1. Choose one of the three services.
  2. Write the proof of willingness to pay in one sentence: who pays, how much, and by when.
  3. Have one paid or prepaid conversation.
  4. Only then start the second service.

Source

Prompted by Greg Isenberg’s post on AI-native services as an opportunity: x.com/gregisenberg. The shortlist and the willingness-to-pay rule are my own. I first shared a short version on X on 21 September 2026.

Also see: Meeting brain: a queryable decision log · Public work