Our Methodology

How we review AI agents

What a review on AI Agent Square is based on, what it is not based on, and how to tell us when we got something wrong.

What a review is based on

The vendor's own material. Pricing pages, plan comparisons, documentation, security and compliance pages, release notes and official announcements.
Independent records. Funding, acquisitions and dates come from filings, the vendor's press releases or major news outlets.
Our own use, where it exists. If Fredrik or Morten has used a product, the review says which plan and what we saw. If not, the review says it is not a hands-on trial.

Every review lists its sources at the end, with the date each one was checked. If we cannot source a claim, we leave it out.

What a review is not

It is not a lab benchmark. We do not publish performance figures we did not measure.
It does not carry a score out of ten. We removed scores in October 2026 because most reviews were not based on testing. A review gives a plain verdict instead: who the product suits, who it does not, and what to check before buying.
It is not paid for. No vendor pays for coverage, placement or wording. We run no ads and no affiliate links.

Corrections and updates

How do I report an error?

Use the contact form and choose "Reporting a factual correction". Include the page, the claim, and a link to the source that shows the right figure. If we were wrong, we correct the page and change its "last reviewed" date.

How current is a review?

Each review shows the date its figures were last checked. Vendors change pricing often, so check the linked source before you buy.

Do you use AI to write reviews?

We use AI tools to help research and draft. A named author checks every figure against its source before publication.

Put It to Work

Browse the reviews

Every review on AI Agent Square is made the way this page describes.

Compare AI Agents Meet the Team