The method

HOW WE SCORE

We celebrate the ones that solve a real problem. We flag it plainly when a claim does not add up. And we share the lesson so you can copy what works.

Some run in the background. Some do the work you buy. We show who does what.

How reveals work

TWO BUTTONS, THREE ANSWERS

You guess Agentic or Human. Sometimes the answer comes back HYBRID. You still only get two buttons. That is the game. Hybrid just shows up at the reveal, when the work is honestly split. Most places will land somewhere in the middle, which is kind of the point.

The two scores

WHO DOES IT, AND IS IT WORTH IT

Human in the loop

How much delivery the team, not the customer, still does, on a scale of 1 to 10. Lower means agents do more of the delivery work.

Agentic sweet spot

Our call on problem, uniqueness, human touch, and money. It comes with a plain answer to one question: should you model it?

Scores are editorial, not paid ratings. Sponsors and paid submissions never change a verdict or a score.

Evidence labels

IF WE ARE GUESSING, WE SAY SO

We tag everything Confirmed, Claimed, or Inferred. If we are guessing, we say so.

  • Confirmed We can see it on their site or in a source we can point to.
  • Claimed They say so, and we have not checked it independently.
  • Inferred Our best read from what is public.
  • To confirm Not known yet. The owner is welcome to tell us.
Stage labels

HOW FAR ALONG IS IT?

This market is new. Low revenue is okay. We label the stage honestly, and we update it when the proof changes.

  • Worth watching Early. Low or pre revenue. Real and live. We are watching if it finds buyers.
  • Getting traction Repeat buyers, sales, or a clear demand signal.
  • Making money Clear money signal. A price people pay, plus sales or repeat use. If it is only claimed, we say claimed.