How we research

Researched means we checked the setup against published sources, but we did not run it end to end ourselves. We read the vendor docs, practitioner writeups, and forum threads behind it, rebuilt the steps from those sources, and linked every source on the page so you can check our work.

If a step only works in some cases, we say so on the page. Time and money figures on Researched pages are labeled estimates, not measured results.

Tested means we ran it ourselves. The page carries a short receipt note saying what we ran and what happened. We only use this sticker when there is a real run or receipt behind it. If we have not run it, it stays Researched, even when the sources look solid.

1. Start from the job, not the tool. We pick a real job someone wants done, then look for the simplest way to do it with AI.

2. Go to primary sources first: vendor documentation, official help pages, and practitioner writeups from people who did the job. Forum threads count as leads, not proof.

3. Cross-check the claims. If two sources disagree, we say so or leave the claim out. We do not invent numbers, quotes, or results.

4. Rebuild the steps so you can copy them. Prompts on the page are copy-paste ready, and they tell you to approve before anything sends or spends.

5. Label what is estimated. Time and money figures are estimates unless the page says we measured them.

6. Re-check pages over time. Pages carry a Reviewed date from the last pass.

We do not claim independent verification of every setup. Researched means checked against sources, not run by us. If a setup stops working, tell us at hello@aisetupsthatwork.com and we will re-check it.