Pre-publish
Where the AI blog post validator sits in your routine
A draft is finished, nothing is scheduled yet, and you want a second read before it goes out. That is the slot this fills. Paste the post with a short brief on top and it returns a written review of the argument and the structure: where the introduction misses the question, which heading covers a section that does not exist, which claim is asserted and never supported.
It runs from the box above. Free, 5 runs per rolling 24 hours, up to 12,000 characters covering the brief and the draft together, no account. Your text goes to our AI server for the review, and there is no URL field, so paste the text rather than pointing at a page.
The brief
Three lines above the draft change the whole review
Paste a bare draft and you get generic editorial notes. Paste these three lines first and the review starts arguing with the post instead of describing it.
- The query
- What someone typed to arrive here. The review can then check whether the opening actually answers it, which is the single most common failure in a draft.
- The reader
- Who this is for and what they already know. It decides whether a section is necessary background or padding.
- The one thing
- What the post has to convince them of. Anything that does not serve it is a candidate for cutting, and the review can say which.
- Optional fourth
- Where the post is going: a newsletter, a docs site, a company blog. Format expectations differ and saying so is cheap.
How to
How to run the AI blog post validator in three steps
One pass for the writing, one pass for the counting. They are different jobs and mixing them is why things get missed.
- Step 1
- Write the three brief lines, paste the draft underneath, and add an instruction if you want a narrow read such as the structure only.
- Step 2
- Act on the review section by section. Fix the thing named rather than rewriting the whole piece around a general comment.
- Step 3
- Run the countable checklist below by hand, then clear invisible characters with the scanner on this site as the last action before publishing.
Do this yourself
The countable checks the AI blog post validator does not do
None of these involve judgement, and none of them happen here. They take a couple of minutes and they are the half of a pre-publish check most drafts fail.
- Title length
- Count the characters. This site holds its own titles to 60 and will not build past it. One checker in this space publishes 40 to 70 as its range.
- Description length
- Same job. This site's audit caps its own descriptions at 160 characters.
- One H1, then H2s in order
- Check the heading levels do not skip. Nothing on this route counts them for you.
- Alt text on every image
- Open the draft and look. An empty attribute is right for a decorative image, a missing one is not.
- Links resolve
- Click every one. No link is ever requested by this tool, so a dead URL passes straight through the review.
- Invisible characters
- Run the local scanner. It covers exactly 14 of them and gives a definite answer in seconds.
What you will not get
No score comes back, and that is deliberate
Other tools in this space return a percentage with bands attached: above 80 is strong, 60 to 79 needs work. Neither of the two we looked at will say whether the number is calculated by rules or written by a model, which is the only thing that would make it meaningful.
This route is a text generator, so a score from it would be generated prose wearing the costume of a measurement. There is a practical problem with grades as well. A number that goes green ends the review, whereas a written note about paragraph six has to be read and either acted on or argued with.
Straight answer
What the review never sees
It sees the text you paste, and nothing else. Nothing on this page opens a URL, so your live page, your CMS rendering, your internal links and your redirects are all invisible to it. A broken link in the draft comes back unmentioned because no link was ever requested.
There is no corpus either, so it does not detect copied or stitched-together passages, and it computes no readability formula, since those are arithmetic on syllables and sentence lengths rather than acts of reading. If you want Flesch Kincaid or a link audit, use something that actually does them, and treat this as the part of the job that needs an opinion.
Last pass
The last thing to do before you hit publish
If any of the post was drafted with a chat assistant, paste it through the hidden character scanner on this site before it goes into the CMS. Zero-width spaces, word joiners, a stranded byte-order mark and no-break spaces all copy across invisibly, and they surface later in slugs, in search, and in whatever tool next reads the file.
Then reread the title and the description as they will appear in a result, not as they look in your editor. Those two strings do more work than any paragraph in the post, and they are the two fields it is easiest to overrun without noticing.
In practice
What a validator cannot promise about ranking
Nothing here predicts search results, and no suggested change is promised to move a page. Word count rules circulate in this space as though they were mechanics: 800 words as a floor, 1,200 to 2,000 for competitive topics. Those are third-party conventions, and nothing on this page verifies them or repeats them as advice.
The same goes for AI detection. This does not tell you whether a draft will pass a classifier, and a round-up of blog quality tools states the position plainly: the classifier it tested is not foolproof and produces both false positives and false negatives. Writing to satisfy one is a poor use of a pre-publish pass.