How we test and score every AI tool
A look under the hood at the rubric behind every score on launched.tools — what we test, how the 100-point scale works, and what gets a tool rejected.
There are thousands of AI tools and almost no honest, consistent way to compare them. Most "top 10" lists are affiliate rankings in disguise. launched.tools exists to fix that: every tool in the directory is reviewed against the same rubric and scored out of 100, and the score never moves because someone paid.
The rubric
Every tool is scored on five public axes, each out of 100:
- Capability — does it actually do the hard part of its job well?
- Usability — onboarding, interface, and how quickly you get value.
- Value — what you get for the plan you pay for.
- Reliability — does it hold up in real use and fail gracefully?
- Docs & support — documentation, examples, and the support surface.
The headline score is a weighted blend of those five, and each axis is shown on the listing so you can see exactly where a tool is strong or weak.
How a review happens
We look at the live product, not just the marketing page — what it claims, what it actually does, how it is priced, and how it compares to the alternatives already in the category. The write-up is specific: real strengths, real trade-offs, and who it is (and is not) for.
What gets rejected
Vaporware landing pages, thin GPT wrappers with nothing of their own, and anything in a prohibited category are turned away. Being listed means a tool cleared the bar — the score then tells you how far past it.
No pay-to-win
Founders can submit and, later, pay for placement or a faster review — but never for a better score. The number is editorial. That separation is the whole point of the directory.
Browse everything we have reviewed on the all tools page, or start with a category.