How AITried Reviews AI Products

AITried’s tagline is “We tried the AI so you don’t have to.” That promise only means something if you can see exactly what “tried” covers on each page. This page explains how we research AI products, how we label the evidence behind each article, how we handle money, and what happens when we get something wrong.

It applies to every review, comparison, guide and “best of” list on the site.

Why we publish our method

Google’s guidance for creators suggests asking “Who, How, and Why” about any piece of content. For product reviews in particular, it says readers build trust when they understand how the review was produced. It also stresses that of the qualities it looks for (experience, expertise, authoritativeness and trustworthiness), trust is the most important.

We agree, and not because of search rankings. AI products change quickly, and their marketing often blurs what a feature really does. A reader deserves to know whether a claim comes from the company’s own documentation, from someone actually using the product, or from neither.

Evidence levels

Every article carries one evidence label near the top. It describes the strongest kind of evidence behind the article as a whole.

Some rules follow from these labels:

• We never describe documentation as experience. A documentation-based article will not say a feature “works well” or “feels natural”. It will say what the company says the feature does, and link to where it says so.
• Hands-on claims will name their scope. When an article moves to hands-on tested, it will state which plan, which platform and which dates the observations cover.
• Labels can change. If a documentation-based article is later backed by hands-on use, we will update the label, the date and the text together.

At launch, all AITried articles are documentation-based. That is deliberate. We would rather publish an accurate, clearly sourced summary than an impression dressed up as a test.

How we source facts

Every material fact in an article, such as a price, a limit, a policy, a feature or a regulatory step, must be supported by a source listed with the article. Our sourcing order is:

1. The company’s own official pages: help centre, pricing page, terms of service, privacy policy, release notes.
2. Platform documentation: Apple, Google Play and similar, for subscription, refund and listing rules.
3. Regulators and public bodies: consumer-protection agencies, data-protection authorities, police fraud services.
4. Reputable reporting, used only for context and never as the sole support for a product claim.

We do not treat forum posts, unofficial wikis, social media screenshots or other review sites as proof of a product fact. Each source is listed with the date we accessed it, because pricing and limits for AI products can change week to week.

What we will not publish

• Star ratings, scores or rankings not backed by stated criteria and evidence.
• User numbers, revenue figures or market-share claims without an official or verifiable source.
• Testimonials, quotes from “users” or invented case studies.
• Discount codes or “exclusive deals” we cannot verify.
• Claims that an AI product improves mental health, relationships or wellbeing.

When we cannot verify something, we say so in the article instead of guessing.

No paid verdicts

Companies cannot pay for a review, a higher position in a list or a more favourable verdict. We do not accept payment to publish, change or remove coverage. If a company contacts us about an article, we treat it like any other correction request: we check the claim against sources and update only if the evidence supports it.

Affiliate policy

AITried may earn a commission when readers buy through some links on the site. That is how an independent publication like this pays for itself. Our rules:

• Every affiliate link will be disclosed. The FTC’s guidance says disclosure should be clear and conspicuous, and that “the closer the disclosure is to your recommendation, the better.” It also says the label “affiliate link” on its own is not adequate. We will use plain wording near the links, explaining that we may earn a commission.
• Commissions never change a verdict. Whether a product has an affiliate programme has no effect on what we write about it, where it appears in a list, or whether we recommend it.
• We cover products without affiliate programmes too, and we will tell you when a free or cheaper option is good enough.
• We do not hide the cancellation route. Every paid product we cover should link to how to cancel. See our guide on how to cancel an AI subscription.

Corrections

We will get things wrong sometimes, often because a company changed a price or a policy after we published. When we find or are told about an error:

• We check the claim against the current official source.
• We fix the text and update the source list and access date.
• For material errors, such as a wrong price, limit or policy, we add a short correction note to the article explaining what changed.
• We do not silently rewrite verdicts.

You can report an error through the contact details on the site. Please include the page, the claim and, if possible, a link to the official source.

Update cadence

AI products move fast, so articles have a planned review schedule:

Each article shows its publication date. When we substantially update one, we will show the update date and refresh source access dates.

How this looks in practice

Take our guide to cancelling an AI subscription. It is labelled documentation-based. Every step comes from Apple, Google or FTC support pages listed in its sources. It does not claim anyone cancelled a particular product, and it tells readers where store rules end and a company’s own terms begin.

Our companion-app pages follow the same pattern. The Candy AI review and Candy AI alternatives pages summarise what is documented, and our framework for evaluating an AI companion app explains the criteria. Practical guides like the AI companion privacy checklist, AI companion subscription costs, free vs paid AI companions, what memory means in AI companion apps and how to spot fake AI app clones give you tools to check claims yourself, including our subscription calculator.

Bottom line

Every AITried article tells you what kind of evidence stands behind it: documentation-based, hands-on tested or not yet tested. Facts are sourced to official pages with access dates. Nobody can pay for a verdict, affiliate links are disclosed and never change what we recommend, errors are corrected openly, and articles are reviewed on a schedule. If a page does not meet these standards, tell us and we will fix it.