A catalog listing is not a claim that a product has been hands-on tested. We distinguish catalog research, original quality reviews, and use-case Fit Scores. No pay-for-placement or affiliate-driven quality ratings.
Catalog research checks identity, advertised capabilities, pricing and source evidence. A Review Score is an original product-quality judgment only when representative testing supports it. A Fit Score remains a reference for a named use case. Unreviewed, limited and stale records are labeled; none is silently treated as a verified quality score.
The AI tools market adds hundreds of new products every week. Most directories list everything that ships and leave you to figure out what is worth your time. We think that defeats the purpose of a directory. Our job is to do the evaluation work so you don't have to.
We do not accept payment for listings. We do not rank tools based on affiliate commission rates. If a tool is listed here, it is because it met our criteria on its own merits. If a tool is not listed, it either failed evaluation, has not been evaluated yet, or was removed after a periodic re-review.
What does the product claim to do, and what evidence supports that claim? Source research establishes advertised capabilities. Actual performance claims require an original review with a stated test method; listing alone does not establish them.
What we look for: accuracy of output, feature completeness relative to claims, edge-case handling, speed of execution.
What is known about consistency, failure handling and recovery? Distinguish documented limitations from observed test results. A listing is not an uptime guarantee, and repeated testing is claimed only when the review record documents it.
What we look for: uptime consistency, output quality variance, API stability (for developer tools), error handling.
Is the pricing fair relative to what you get? We evaluate free tiers honestly, noting what is actually usable versus what is a trial with a different label. For paid tools, we assess whether the output justifies the cost compared to alternatives in the same category.
What we look for: free-tier limitations, price-to-capability ratio, usage caps, hidden costs (API overages, seat fees).
Does it fit into real workflows? A tool that works in isolation but cannot connect to anything else has limited practical value. We prioritize tools that play well with others: through APIs, native integrations, Zapier/Make connectors, or standard file formats.
What we look for: API availability, native integrations, export options, automation platform support, data portability.
Is it actually different from the other options in its category? The AI tools market is full of products that are thin wrappers on the same foundation models with different branding. If a tool does not offer a meaningfully different experience, capability, or price point from existing listings in the same category, it does not get added.
What we look for: unique capabilities, proprietary models or data, distinct UX approach, category-defining features.
Tools enter our evaluation pipeline through three channels:
Listings are not permanent. Tools are removed or flagged when:
Material changes to a product, plan or source evidence trigger reassessment. Read the pricing-check, test and publication dates separately; a catalog update does not mean a product was retested. Evidence that needs refreshing is marked rather than silently treated as current.
Review history and prior evidence are retained. A new quality score requires supported observations and an explicit editorial judgment. Unknown legacy ratings and external references are not converted into new Review Scores.
Questions about our methodology or want to suggest an improvement? Get in touch.