How we review
TaskFirst AI is a meta-review site. We do not pretend to have spent months living with every tool. Instead, every guide is built the same way, in this order, and we tell you exactly which claims are researched and which are tested.
1. We aggregate credible reviews
We read what named publications (TechRadar, Tom's Guide, The Verge and similar) and real users (Reddit, Discord and other forums) have said in roughly the last six months, and we cite them by name. If we write “TechRadar found X” or “the Reddit consensus in r/[community] is Y”, that is a real, attributed source — not our own invented experience.
2. We verify pricing ourselves
We check current pricing in GBP directly against each vendor's own pricing page, including free-tier limits and commercial-use terms. Every comparison table carries a “last verified” date that is separate from the article's publish date, so you can see how fresh the numbers are.
3. We do small, honest first-hand checks
Where we can, we run a like-for-like task through each tool's free tier — for example the same short script through several voiceover tools — and link the actual outputs. These are quick touch-tests, and we describe them as exactly that. We never fabricate long-term-use anecdotes to sound authentic.
Researched vs. tested
Every article is labelled. Researched means we read and verified sources and pricing but did not run the tool ourselves for that piece. Tested means we used it, and the article includes the evidence — what we ran, and links to the output. We only use the word “tested” when that evidence is present.
We commit, and commercials come last
Each guide ends with a named recommendation and a one-sentence reason — no “worth considering” hedging. We decide the editorial verdict before we look at commission rates, so affiliate availability never changes who we recommend. We do not publish invented ratings, fake review markup, or first-person claims about tools we have not used.