Editorial Methodology

How AI tools are selected

The AI Tools catalog is a small, manually curated list, not an automated directory — each entry is added individually after research, not scraped in bulk. Tools are chosen for having a genuine, checkable free tier or a clearly stated paid-only status, a distinct real-world use, and enough public documentation to verify claims against. We do not accept payment, free credits, or any other consideration in exchange for a listing or a favorable description.

How prices and plans are verified

Every price, free-plan limit, and feature claim on an AI-tool page is checked directly against that provider's own official pricing page at the time of writing, then cross-referenced against at least one independent secondary source where possible. The exact page checked, its type (official documentation vs. a secondary source), and the date it was checked are recorded and shown in the "Verification" box on each AI-tool page — never presented as a fixed fact with no origin.

Each entry also carries a reviewStatus (verified, needs-review, or deprecated) and a nextReviewAt date, roughly 90 days out from the last check. Pricing changes — a page that hasn't been re-checked past its review date is flagged as such, not silently presented as current.

How workflows are tested

Each entry in Workflows combines an AI tool with one or more of our own browser-based tools into a specific, described sequence. Every step explicitly states whether it processes data locally (in your browser, nothing uploaded) or sends data to an external AI provider — this distinction is checked against each tool's actual behavior, not assumed. Quality checks and failure-recovery notes on each workflow page describe specific, known ways the sequence can go wrong, not generic disclaimers.

How prompts are evaluated

Each entry in Prompts lists which AI tools it's meant for, a quality checklist for judging the output, and known failure modes for that specific task. Example outputs shown on prompt pages are explicitly labeled illustrative — none of them is a captured, verbatim response from a specific model, since a model's real output varies by version, context and run, and presenting one as fixed would misrepresent that variability as reliable.

How calculators are verified

Calculators that depend on a country's tax code, labor law, or official rates (currently France and Morocco) display their jurisdiction, the applicable year, the official legal source, and the date that source was last checked, directly on the tool page — not buried in a footnote. The underlying formulas are covered by automated tests against known, worked examples from the official source, run before every deployment. A calculator's FAQ section states explicitly what it does and does not cover (for example, professional categories with non-standard deduction rates), rather than presenting a general rule as universal.

How sources are cited

A source is either the provider's own official page (pricing page, product documentation, or — for legal/tax content — the official government or legislative text) or, where that isn't sufficient alone, a secondary source used to cross-check it. Sources are linked directly from the page that uses them, not just referenced by name. We do not fabricate prices, ratings, review counts, testimonials, or dates — where a detail couldn't be independently verified beyond what a provider publishes, the page says so explicitly rather than presenting a guess as fact.

How errors are corrected

If something is wrong — a stale price, a broken link, a miscalculated example — the fix is made directly on the affected page, not left standing behind a general disclaimer. Corrections are treated as normal maintenance, not an exceptional event requiring a visible retraction notice, since nothing here is presented with a false claim of permanence in the first place.

How pages are updated, and how outdated information is handled

Every tool, guide, AI-tool listing, comparison, alternative, workflow and prompt page carries a real, hand-tracked date (verification date, review date, or last-updated date depending on the content type) — these dates are never recomputed automatically on every deploy, so a date on the site always reflects when that specific content actually last changed. An AI-tool entry whose nextReviewAt date has passed is marked needs-review rather than continuing to display as freshly verified.

How to reach us

To report an error, a stale price, a broken link, or to suggest a tool or correction, use the Contact page. See also About Everyday Tools for who operates the site, and the Privacy Policy for how data is (and isn't) handled.