Editorial Standards
Last Updated: September 3, 2026 · Accountable author: William Vasquez
The Short Version
Posts are AI-drafted and machine-validated before publish: markup stripped, internal links checked against the real route table and dropped if they do not resolve, moved source URLs repaired, duplicate headlines rewritten, and a 100-point rubric enforced with an 85 minimum. Those checks verify structure and link integrity — they do not independently verify every statistic or quotation. Follow the citations, and tell us when we are wrong.
Table of Contents
This page describes the checks a blog post passes before it appears on /blog or /es/blog, and — just as importantly — the checks it does not pass. It is a companion to our editorial policy, which explains who is accountable and how posts are produced.
Start from the same disclosure we lead with everywhere: our posts are AI-drafted and no human editor reads them before publication. What stands between a generated draft and a published page is a set of automated validators and a scoring gate. Those catch a specific and knowable class of problems. They do not make an article true, and we are not going to pretend otherwise.
Every draft passes through the following, in order, on each of the three daily runs. None of these are aspirational; they are code that runs on every article in both languages.
Any HTML the model emits inside body text — tags, inline styles, entities — is removed, because those fields render as plain text and stray markup shows up as literal characters on the page. Blocks that contain nothing but a decorative separator are dropped.
The site's actual route list is derived from the filesystem — every directory holding a page or route file — rather than guessed. Any internal link the model invents (a plausible-sounding page that does not exist) is removed and its anchor text is unwrapped back into ordinary prose. This is why you will not find a broken in-site link in a post: unresolvable links never publish, rather than being redirected after the fact.
We keep a hand-verified list of cited source URLs that publishers have relocated, and rewrite them at publish time so citations land on a live page instead of a 404.
Before saving, the draft's title is compared against the title of every existing post. On a collision, the headline is rebuilt from the post's own slug — which already describes that specific article — rather than published as a second copy of someone else's headline.
A second H1 inside body content is demoted to H2 so each page has exactly one top-level heading. Malformed image paths are repaired. Over-long meta titles and descriptions are truncated to fit search results.
Each candidate news story is fingerprinted and compared against every post already published, so the pipeline does not write the same article twice under a different headline.
After cleaning, the draft is scored against a twelve-criterion rubric worth 100 points: title, meta description, length, keyword usage, heading structure, internal links, external links, FAQ coverage, calls to action, table of contents, and legal citations. Below 85 out of 100, the draft is sent back to the model for refinement rounds and re-scored.
Four conditions fail a draft outright regardless of its score: fewer than 1,000 words, no FAQ section, no calls to action, or a title longer than 65 characters. The published score for each post is stored alongside it in our database.
Be clear about what this rubric measures. It measures whether an article is complete, well-structured, properly cited in form, and genuinely substantive in length. It does not measure whether the article is correct. A well-structured piece with a wrong statistic in it can score 92.
This is the section most pages like this one leave out. Ours is explicit, because a policy that overstates its own rigour is worse than no policy.
When an article says a firm raised a specific sum, that a court ruled a particular way, or that a survey found a given percentage, that claim comes from the drafting model's reading of its sources. No person confirms each one against the primary document before publication. Treat figures, dates, and dollar amounts as reported rather than as verified, and follow the citation.
Our brief instructs the model to reference the actual statements of named, real people. We do not separately authenticate the wording of a quote. If a quote attributed to a named individual is inaccurate, that is exactly the kind of error we want reported, and we will correct it.
We repair moved URLs. We do not audit whether a cited outlet got the underlying story right.
The four images on each post are AI-generated. They do not depict real events, real premises, or real people, even where the article discusses named individuals.
Spanish posts are translated from the English article in the same automated run, not independently researched. A factual error in the English text will appear in the Spanish text, and a correction to one is applied to both.
You do not have to take any of this on trust. Three things you can do in under a minute:
Follow the external citations in a post and confirm the figure appears in the source. Check the byline and the author profile at /about/william-vasquez against the public bar record. Compare the post's last-modified date against the correction history described in our corrections policy.
If a claim in one of our articles is load-bearing for a decision you are about to make in a legal matter, verify it against the primary source and against a lawyer licensed in your jurisdiction. Nothing here is legal advice, and reading it creates no attorney-client relationship.
William Vasquez is the named author on every post and the person answerable for its accuracy. He is a practising attorney admitted to the North Carolina State Bar since 2011, admitted to the U.S. Courts of Appeals for the Fourth, Fifth, and Eleventh Circuits, a member of the American Immigration Lawyers Association, and has handled 3,000+ matters personally across immigration, personal injury, criminal defence, family law, and workers' compensation. He holds a B.S. in Computer Science from Campbell University and a Master of Divinity from Oral Roberts University, and served seven years in the U.S. Air Force as a Defense Intelligence Agency Spanish linguist.
Accountability in this operation is post-publication. It is not a claim that he pre-approved each article; it is a commitment that errors reported against his byline get fixed and recorded.
Use our contact form or email info@hodos360.ai. The more of this you can include, the faster it gets fixed:
The URL of the article. The exact sentence or figure you are disputing. What you believe the correct information is, and a source for it if you have one.
Every report is read. What happens next — whether the post is corrected in place, rewritten, or removed — is set out in our corrections policy.
When we add, remove, or change a pre-publication check, this page is updated and the "Last Updated" date above changes with it. If we ever introduce human fact-checking before publication, it will be described here — and until it is, assume the checks listed above are the whole of it.
Send the URL, the sentence, and your source. We will check it.
Contact form · Editorial policy · Corrections policy
Email: info@hodos360.ai
Hodos360 PLLC · Metro Office Park 2, Calle 1, Suite 100, Guaynabo, Puerto Rico 00968