Editorial standards
Nutron publishes one article a week about business operations, process automation and custom software. This page describes exactly how those articles are made, because anyone deciding whether to trust a number on this site is entitled to know where it came from.
How an article gets made
- Subject. Topics rotate through a fixed set of subject areas so the archive covers the whole of what Nutron does, rather than whatever is loudest that month. Roughly one article in four is angled at the Phoenix and Arizona market, because that is where Nutron is.
- Research. A grounded search establishes what has actually happened in that subject area recently, and gathers the specific claims and figures the article will rest on, each attached to the page that substantiates it.
- Source verification. Every candidate source URL is fetched. Links that fail, time out, or silently redirect to a homepage are dropped. If fewer than five sources survive, the article is not written.
- Drafting. The article is drafted from the verified source set only. It may not cite a URL that was not verified, and it may not link to a Nutron page that does not exist.
- Automated checks. The draft is rejected if it is too short, carries fewer than five outbound citations, contains a figure in a paragraph with no citation, links anywhere it should not, or uses the marketing register we have banned.
- It is saved as a draft. Not as a page. A draft is absent from this blog's index, from the sitemap and from llms.txt, so nothing links to it, no search engine or answer engine can reach it, and it has no public URL. It sits there until somebody deals with it.
- A person publishes it, or does not. Someone reads the draft, checks the claims against the sources, and adds something the model could not know — a delivery detail, a decision and why it was made, a failure we have hit. If there is nothing to add, the draft is deleted rather than published. Skipping a week is an allowed outcome and happens. Whoever publishes is recorded as the reviewer and named at the foot of the article.
The order of those last two steps is the whole point. A machine can research and draft; it cannot decide that a piece of writing is worth someone's attention. Publishing is the one step we have deliberately not automated.
Where AI is used, and where it is not
AI is used for research, drafting and the photograph at the top of each article. It is not used to invent statistics, quotations, customers or case studies, it does not decide what gets published, and it is not given a byline. Articles are attributed to Nutron Engineering — the team accountable for the content — rather than to an individual who did not write them. Generated images carry Google's SynthID watermark.
Earlier articles on this blog carried invented author names. They did not correspond to real people, so they have been replaced with the Nutron Engineering byline and the pipeline now refuses to publish a post attributed to a person without a real profile behind them.
Every article's footer names the models that drafted it and the date it was last reviewed.
Sourcing
- Any number that is not arithmetic shown in the article is attributed in the sentence that uses it, naming the source and the year, and linked to that source.
- Nutron's own results are described as client-reported and linked to the case study that substantiates them. We do not publish a Nutron figure without the engagement behind it.
- We prefer primary sources — the research report, the filing, the regulation, the vendor's own documentation — over coverage of them.
- Quotations carry a named speaker or a named institution, or they are cut.
Corrections
If something here is wrong, tell us and we will fix it. Email support@nutron.ai with the URL and the claim you are disputing. Corrections are made to the article itself and the revision date is updated; we do not quietly delete pages that turn out to be wrong.
What we do not do
- No sponsored posts, guest posts or paid link placements.
- No city or industry permutations of the same article. If one skeleton would serve two posts, it is one post.
- No changing a publication date to make an article look fresh. Revision dates change only when the content actually changes.
- No statistic without a source. No source without a link that resolves.
Questions
- Are Nutron blog articles written by AI?
- Drafts are. Each week a pipeline researches a current development using Gemini with Google Search grounding, then drafts the article. Every source URL is fetched and confirmed to resolve before the draft is written. Nothing publishes automatically: the draft is invisible to the site until a named person reads it, checks it and publishes it. Articles are bylined to Nutron Engineering rather than to an individual, because no individual wrote them.
- Does anything publish to this site without a human reading it?
- No. Generated articles are written as drafts. A draft is absent from the blog index, the sitemap and llms.txt, so it is never linked, never crawled and has no public URL. It becomes a page only when a person publishes it, and that person is recorded as the reviewer and named in the disclosure line at the foot of the article.
- How are statistics checked?
- Every number in an article must be attributed in the sentence that uses it and linked to the primary source it came from. The build rejects a draft that contains a figure in a block with no citation link, and rejects any outbound link that was not fetched and confirmed live during generation. Numbers that cannot be sourced are removed rather than softened.
- What happens if an article is wrong?
- Email support@nutron.ai with the URL and the specific claim. Corrections are made to the article itself, the revision date is updated, and material corrections are noted in the article. We would rather fix it than defend it.
- Does Nutron publish sponsored or guest content?
- No. Everything on this blog is written for Nutron, about work Nutron does. We do not accept paid placements, guest posts, or link insertions.