Editorial standards
Wording revised
Every post here is produced by software and AI models. No person writes it, edits it or signs it off. This page lays out the steps and where they fall short.
The making of a post
An AI model that runs on our own local hardware writes each post from start to finish. Software then attaches every statement to a source and prints that source on the page. Last, a fact-checking model from an outside AI firm looks the post over, and it does so a single time. Nobody types, polishes or signs off a post.
How sources are used
Each factual claim sits next to a source you can open, and the sources are listed again below the post. For product behavior we lean on the vendor's own documentation. Where a page names the plan, the version or the permissions a task needs, the post repeats them. Every post prints the day its sources were last looked at.
Help pages and real tasks
Reading a help page is a different act from trying the task in a signed-in product. A post says which one it relied on. No post reports a test in a signed-in account.
Where the review stops
The review model can miss an error. We do not present it as expert review or human oversight. Treat each post as a starting point and open the sources before you act.
Examples and numbers
Where we use an example, it is labelled as one. Nothing on this site is invented: no figures, people, awards, client names or reader counts.
Scores and methods
A health score is an indicator. It is not a measured chance that a customer will cancel, and its weights depend on the account. Nothing on this site guarantees retention.
Legal questions
Posts give general operational information, not legal advice (notice periods or renewal clauses). Your own legal team should read your agreements.
If you think a post is wrong
Follow the steps on the Corrections page.