Editorial policy and methodology

Selection and scope

The site starts with the subjects supplied for review, then checks whether each one actually belongs in the intended category. Candy AI, Lovescape, GoLove AI, and CrushOn.AI had enough relevant evidence for compact comparison. JoyAI did not receive a ranked placement because the reviewed category evidence remained ambiguous.

Comparison dimensions

Testing method

Supplied routes are opened read-only and classified by their observed outcome. No account, message, payment, age confirmation, or external form is submitted merely to manufacture evidence. Public product materials are preferred for feature, price, and policy claims. When public sources conflict, the conflict is part of the answer.

Editorial persona selection and the prompt workshop are local site interactions, not tests of an external model. The site never uses its own fictional characters as evidence that a destination contains those characters.

Independent review rules

Compensated outbound links do not justify a higher rating, false endorsement, fabricated scarcity, or omission of relevant limitations. Every named service is evaluated under the same evidence rules, and an outbound action appears only when the observed destination matches its label.

What we do not publish

• Invented star ratings, market-share figures, user counts, testimonials, or universal “best” claims.
• Unverified promotions, countdowns, expiring-price claims, or pressure language.
• Exact feature or price promises derived only from old screenshots or third-party summaries.
• Claims that an editorial image or persona represents an external account.

Corrections

A supported correction should update all affected formats together: HTML, Markdown, JSON, structured data, navigation, and sitemap metadata. A public correction channel will be added only after a monitored recipient is configured and verified.