How We Test AI Tools
Every tool in the AiRecMark index gets a structured record: pricing, plans, features and a five-dimension score built from published documentation and our own checks. Tools selected for an in-depth review are used on real work for at least two weeks before we publish a verdict. We do not republish vendor benchmarks, and no vendor can pay for a better result.
From shortlist to published review
Choose what to test
We pick tools readers actually ask about, plus new releases that matter in each category. Reader submissions are considered, never guaranteed.
Use it for real work
In-depth reviews run at least two weeks on paid or trial accounts: writing, coding, research, design, video or ops tasks that match how readers use it.
Score five dimensions
Each dimension gets a score out of 100 from our own testing notes. The overall score is a weighted composite, not an average of vendor marketing.
Publish, then revisit
Scores and starting prices are re-checked quarterly, and sooner when a vendor ships a major model, plan or pricing change.
Five scoring dimensions
Features
Does it do the job it claims, and how deep does it go beyond the demo?
Value for money
What you get at the entry price, and what forces an upgrade.
User experience
Time to first useful result, learning curve and day-to-day friction.
Output quality
How good the actual output is on the tasks readers care about.
Support & docs
Documentation quality, onboarding and how fast support answers.
What we don't do
- blockNo paid placement, score adjustments or queue priority.
- blockNo payment for a positive review, ever.
- blockVendors do not see a score before publication.
- blockNo vendor benchmark presented as our own hands-on result.
Found something out of date?
AI tools change fast: a price, a plan or a model can shift between quarterly re-checks. Tell us which page looks wrong and we will re-test and correct it.
mail [email protected]