AI Tool Recap exists to help you choose the right AI tool without drowning in launch hype. This page explains exactly how we evaluate tools, what our scores mean, what we will and will not claim, and how affiliate revenue stays separate from rankings. If a review on this site does not match what you read here, that is a bug – email our team and we will fix it.
How We Review AI Tools (And What We Don’t Do)
Most “best AI tools” lists are vendor-friendly affiliate charts dressed up as reviews. We do something different, and we are upfront about its limits.
What we do: deep desk research on every tool we cover. We read official documentation, pricing pages, changelogs, security and privacy disclosures, the vendor’s recent product announcements, third-party comparison data, and verified user reviews from G2, Capterra, Reddit, and developer forums. Where a free tier or demo exists, we run it through one or two real workflows before publishing.
What we do not do: long-running hands-on tests where one editor uses a tool for a full project. We are a small editorial team – we cannot honestly run multi-week tests across hundreds of tools at the depth a single specialist publication might run on one tool a month. So we never claim we did. Every review carries a tested-status label that tells you exactly what kind of evidence the verdict is based on.
Tested-Status Labels
Every tool we cover carries one of three labels. Look for the badge at the top of any review or in the quick-picks table on any list page.
- Demo-tested – we ran the tool through a free tier, public demo, or short trial on at least one realistic task. Strong basis for ease-of-use and free-plan-usefulness claims; limited basis for long-term reliability claims.
- Researched – the default label. The verdict is based on official documentation, pricing pages, recent changelogs, third-party comparisons, and verified user reviews. Strong basis for feature, pricing, integrations, and privacy claims; we will not pretend to have run the tool ourselves.
- User-review summary – community signal only. We aggregate G2, Capterra, Reddit, and developer-forum feedback. Useful when a tool is too new, too niche, or too closed for documentation-based analysis. The lowest confidence tier; we say so on the page.
We do not have a “hands-on” or “long-term test” tier. If we ever start running multi-week hands-on reviews, we will add the tier here first and only then start using it on individual reviews.
The AI Tool Recap Scorecard
Where we score a tool, we score it on six dimensions plus a weighted overall. Each dimension is scored 1.0 to 5.0 in 0.1 increments and weighted equally.
- Workflow Fit – does this tool actually fit a real task, or does it just generate output that needs heavy rework? The most important dimension for our editorial angle: tools rank for workflow fit, not feature count.
- Output Quality – accuracy, usefulness, and whether the output survives a critical read. We look at evaluation benchmarks where they exist and at user-reported quality issues where they do not.
- Ease of Use – speed to first useful result, friction in the day-one experience, learning curve. Includes UI clarity, onboarding, and how forgiving the tool is when a prompt or input is imperfect.
- Value & Pricing Clarity – free-plan usefulness, hidden message and seat limits, whether the price is honest about what you get, and whether the upgrade tiers make sense. Sites that hide pricing or surprise-charge lose points here.
- Integrations & Ecosystem – where the tool plugs into the rest of your stack, API and webhook quality, and whether it plays nicely with Zapier, Make, Notion, Slack, and the major cloud suites.
- Privacy & Team Readiness – how your prompts and outputs are handled, whether training opt-out exists, SOC 2 or equivalent status, admin and SSO controls, and whether the tool is deployable beyond one user.
The overall score is the simple average of the six dimensions, rounded to one decimal. Rounding ties go down, not up.
Example Scorecard (Illustrative)
Here is what a finished scorecard looks like on a real review. The numbers below are illustrative – they describe a fictional all-purpose chatbot to show the format, not a real product.
| Dimension | Score | One-line reason |
|---|---|---|
| Workflow Fit | 4.7 / 5 | Strong fit for daily research, drafting, and quick analysis. |
| Output Quality | 4.8 / 5 | High reliability on reasoning and writing; occasional confident errors on niche facts. |
| Ease of Use | 4.6 / 5 | Clean UI; minimal onboarding; mobile and desktop parity. |
| Value & Pricing Clarity | 4.5 / 5 | Generous free tier; one paid tier with transparent message limits. |
| Integrations & Ecosystem | 4.3 / 5 | API is mature; growing list of native integrations; Zapier support. |
| Privacy & Team Readiness | 4.2 / 5 | SOC 2 Type II; training opt-out on paid plans; team controls on the Business tier. |
| Overall Score | 4.5 / 5 | Average of the six dimensions, rounded down on ties. |
Which Tools Get Scored?
We only publish a numeric scorecard on tools we have either Demo-tested or where the documentation and user-review evidence is strong enough to defend a score with confidence. A Researched-only tool can absolutely earn a scorecard – the difference is in how we describe the evidence, not whether we are willing to commit to numbers.
If a tool is too new, too niche, or too closed to defend numeric scores, we will not invent them. The review will show qualitative best-for and skip-if verdicts, a tested-status of User-review summary, and the reasoning behind the verdict in plain English. You will see this most often on early-stage tools and tools that pre-announce features that have not shipped.
Pricing Freshness and Updates
Every review and every list page on this site shows two dates: Pricing checked and Last updated. The first is the date we last verified the tool’s pricing against the vendor’s pricing page. The second is the date we last touched the review for any reason.
Our refresh rules:
- List and hub pages: pricing re-checked at least monthly, and on every notable vendor pricing change.
- Individual reviews: pricing re-checked at least monthly, and immediately on any pricing or product-tier change.
- Scorecards: re-scored on any major model release, major feature shift, or substantive pricing change.
If you see a Pricing-checked date more than 60 days old on any page, the page is overdue and we want to know. Email our team and we will refresh it.
Editorial Independence and Affiliate Revenue
AI Tool Recap earns money primarily through affiliate commissions on tools we recommend. This is a clear conflict of interest if it is not managed transparently, so we manage it with hard rules.
- Editorial picks and sponsored placements are visually separated. Sponsored placements are labelled Sponsored at the top of the section. Editorial picks are not.
- Affiliate links are tagged with
rel="sponsored"per Google’s guidance. - An affiliate relationship does not buy a higher score. If a vendor stops offering an affiliate program, our existing score does not change. If a non-affiliate tool would score higher than an affiliate tool in the same category, the non-affiliate tool wins the spot.
- We disclose affiliate relationships visibly on every page that contains an affiliate link, and in the dedicated affiliate disclosure page.
Corrections and Feedback
If a fact on this site is wrong, a price is out of date, a tested-status label looks inflated, or a tool we ranked is no longer the right call, we want to know. Email our team. Verified corrections get applied within seven days; significant rankings changes get a visible Updated note on the page.