Every review on AI Deep Hub tells you which evidence it stands on. No review here pretends to more than it has.
Why we publish this page at all
Search for any AI tool and you’ll get a wall of articles claiming they “tested” it. Most didn’t. They rewrote the vendor’s feature page, added a star rating, and dropped in an affiliate link. You can’t tell the difference from the outside, which is exactly why those articles work.
We’d rather lose the comparison than win it dishonestly. So here’s our evidence system, in full, including where it’s weaker than we’d like.
What are the two evidence tiers?
Tier A — Hands-on
We created an account, paid for it where a paid tier was needed, and pushed a real task through the tool. Not a demo prompt. An actual job we needed done.
A Tier A review includes:
- A dated screenshot from our own account, not the vendor’s marketing site
- What the task was and what came out of it, including the parts that broke
- The invoice reality: what we were actually charged, including credits and seat minimums
- Whether we kept paying after the first month, and why or why not
Tier B — Documented
We don’t hold an account for this tool. Instead we verify every checkable fact at its source and gather what people who do use it report.
A Tier B review is built from:
- The vendor’s live pricing page, captured as a dated screenshot. Not quoted from another blog. AI pricing changes monthly, and secondhand prices are the single most common error in this niche.
- Official documentation, changelogs, and stated limits (character caps, minute caps, seat minimums, export restrictions)
- Independent user reports at volume: Trustpilot, G2, Reddit threads, and named reviews from publications like PCMag
- Anything material and public that the vendor would rather you didn’t see, including lawsuits, outages, licensing disputes, and policy changes
What a Tier B review will never do is describe an experience we didn’t have. If you read “it feels sluggish on long documents” here, someone had that experience and we’ll tell you who.
How do we verify pricing?
Pricing is where these articles fail most often, and it’s the number that costs you money when it’s wrong.
Our rule: we open the vendor’s own pricing page, capture it as an image with a date on it, and publish that image inside the review. If the price changes next month, you can see exactly when our figure was true.
We also try to publish the number that’s usually missing:
- Annual versus monthly. Many tools advertise the annual-billing price as if it were the monthly one. The gap is often 40% or more.
- Seat minimums. A “$15 per user” plan that requires two seats costs $30.
- Credits. A plan is only cheap until you learn how fast the included credits burn.
- What’s genuinely free. Free tiers that watermark output, cap resolution, or limit lifetime uploads are not free for the job you’re trying to do.
Where a vendor hides pricing behind a login or a sales call, we say so plainly instead of repeating a number we can’t check.
How do we score tools?
We don’t publish a numeric score unless it’s a Tier A review, because a number implies a measurement, and we’re not going to imply one we didn’t take.
Tier B reviews end with a plain-language verdict instead: who this is clearly right for, who should skip it, and what the cheaper or free alternative is when one genuinely wins. That’s the part you came for anyway.
How does affiliate money affect any of this?
AI Deep Hub earns commission when a reader signs up through some of our links. Those links carry rel="sponsored" and there’s a disclosure at the top of every review. Full detail is on our affiliate disclosure page.
Three rules keep the money away from the verdict:
- We keep tools we earn nothing from in the comparison whenever they’re the better pick. Several of our top recommendations pay us zero.
- Ranking order is never sold, and no vendor sees a review before it publishes.
- A negative verdict on a tool we have an affiliate relationship with stays negative. If we can’t say something honestly, we drop the tool rather than soften it.
When do reviews get updated?
AI tools change faster than almost any other software category. A review from eight months ago can be wrong about price, model version, and plan names all at once.
We re-verify pricing and plan structure on a rolling basis, and we stamp each review with the date its facts were last checked. When a fact changes materially, we update the article rather than publish a new one, so the correction reaches the people already reading it.
How do I report something wrong?
Please do. Wrong prices are the most common problem and the easiest to fix.
Email [email protected] with the page and what’s wrong. If you’re right, we fix it and note the correction date on the article. If you work for a vendor and we’ve got your pricing or limits wrong, we especially want to hear from you, though a correction won’t buy a better verdict.
Who writes these reviews?
One person is accountable for everything published here: My Bui, founder and operator of AI Deep Hub. There’s no anonymous content team. You can read her background, contact details, and business address on the about page.
This methodology page was last updated on 10 August 2026. When our process changes, this page changes first.