Editorial Policy
How Dwayo AI writes, sources, dates, and corrects the articles, benchmarks, and case studies on dwayo.ai.
Last updated: October 23, 2026
What this covers
This policy covers the articles, benchmarks, case studies, and free engineering tools published on dwayo.ai. It describes how we decide what to publish, what we claim, and what we do when we get something wrong.
Who is accountable
Every article carries a byline. Dwayo AI is accountable for everything published on this site, whoever's name is on it. Our articles come from work we do: building, evaluating, and operating production AI systems.
How we handle numbers and claims
When we publish a figure from our own measurements, we say what was measured, on what data, on what date, and what the figure does not show. Where we benchmark models, we state the settings used, the number of examples, and the limits of the result, and we publish the harness so the numbers can be rerun. Where we did not test something, we say so. We do not publish a comparison we have not run.
Prices, model names, and vendor features change. A figure tied to a date describes that date only.
Illustrative scenarios and client work
Case studies marked "illustrative" are composite engineering scenarios that show a failure mode and its fix. They are not named client engagements. Where a case study describes real client work, client details are anonymised or withheld where an NDA applies.
Dates and updates
Each article shows its publication date. We change an article's modified date only when its content materially changes, not for formatting or minor edits. We do not refresh dates to make older content look current.
Corrections
If you find an error, email [email protected] with the page and what is wrong. We correct factual errors in place and, for a material correction, note the change on the page.
Commercial relationship
Dwayo AI sells AI engineering services, so articles often link to related services and to our free tools. Read them with that in mind. Technical content is written to stand on its own, and where we recommend an approach we say what it costs and where it does not fit.
Contact
Questions about our content: [email protected]. For how we handle data, see our privacy policy; for how we protect client systems, see our trust and safety framework.