Our review process is designed for a directory that covers many fast-changing adult AI products without pretending that every service has received a paid, long-term laboratory test. The evidence level on the page determines how confident the conclusion can be.
1. Confirm the destination
We check whether the listed URL resolves, where it redirects, whether the product name is still visible, and whether the destination has become parked, unavailable, or unrelated. Duplicate domains that resolve to the same product are consolidated into one canonical page. A failed or mismatched destination is flagged and is not treated as a working recommendation.
2. Record product-specific evidence
We look for the actual workflow, supported media, character or editing controls, platform availability, public policy links, support routes, and limitations visible on the official site. A screenshot is used only when it shows a meaningful part of that destination, such as the homepage, product interface, feature explanation, or pricing page. Decorative logos are not counted as review screenshots.
3. Examine price and commitment
When prices are public, the review records the currency, billing interval, visible entry price, and whether credits or premium actions appear separate from the subscription. We look for renewal, cancellation, refund, and usage-limit information. If checkout or account access is required, the page says that the current total must be confirmed rather than inventing a number.
4. Evaluate the decision, not just the feature list
We assess category fit, useful control, evidence quality, pricing clarity, destination health, privacy signals, and consent risk. The written verdict identifies a suitable user, a meaningful strength, a practical limitation, and the safest next step. Products with thin or inaccessible evidence remain research briefs or are kept out of search until the page can answer a real comparison question.
5. State the limits of the review
A public-site check cannot verify backend deletion, model-training practices, staff access, generated-output consistency, or customer support over time. We do not describe registration, payment, generation, cancellation, or deletion as tested unless those actions were actually completed. Readers should recheck live policies before sharing sensitive data.
Ranking principles
Category order considers relevance to the task, product availability, breadth and usefulness of controls, quality of public evidence, pricing clarity, and safety or privacy signals. Referral availability is not a ranking criterion. Rankings are comparative guidance, not a permanent score, and may change when the underlying product or evidence changes.
How to challenge a review
If you find a wrong destination, outdated price, unsupported statement, or missing disclosure, send the page URL and primary evidence through Contact. Specific evidence is more useful than a request for a higher ranking.