Content Tool

Give us a page and the questions you want it to answer. We score it against what actually wins AI citations, go and find real published sources, and hand back anchored edits - each one named to the research behind it.

Enhance a page What we score on

New here? What this does, and what it refuses to do

This is an enhancement engine, never a blank-page generator. It reads the page you already have, decides where it falls short against eight published factors, and writes patches against text that is already there. Every patch carries an anchor: the exact words on your page that locate where it goes.

The part that costs us the most is the part most tools skip. When a page needs a statistic or a quote, we search, fetch the source pages through our own crawler, and then check that the wording we are about to hand you really exists on the page we are attributing it to. Anything that fails that check is dropped. That is why a run can come back saying we could not verify a source for this gap rather than inventing a number.

We also do not sell you the things that do not work: keyword stuffing measured worse than doing nothing, llms.txt is almost never fetched, and reformatting a page that does not answer the question changes nothing.

The page and the questions

What should we work on? One page, not a whole site. We read it the way an AI crawler does - scripts and navigation stripped - so what gets scored is what a model would actually see.

Page URL

Your content Paste the page's HTML or its plain text. HTML is better where you have it - we can read your JSON-LD dates out of it, which is what a freshness patch needs. Use this for pages behind bot protection or a login, and for drafts that aren't published yet.

What we can't check on pasted content

One factor goes unscored. We never see how your server delivers the page, so we can't tell whether its text is in the HTML a crawler receives or assembled afterwards in the browser. That factor is marked not applicable rather than passed - it still matters on the live page, and it is the one thing pasting cannot tell us. Run the URL once it's published and we'll check it.

Everything else works exactly as it does for a URL, including dates read out of pasted JSON-LD.

Your domain Optional If you tell us whose page this is, we keep that domain out of your source list - a site cannot be its own external corroboration. Leave it blank and we simply won't exclude anything.

Study Optional Attaching a study lets us offer its tracked queries below, and records where those queries stood today so a later before-and-after has a before.

Source pages to read The hard cap on how many external pages we fetch and mine for citable material in one run. We tell you how many candidates we found, so you always know what the cap left out.

Target queries 2 to 6 The questions you want this page to be the answer to, worded the way a person would ask them. Relevance to the actual wording outweighs every stylistic fix in every controlled study, so these drive the whole run.

Tone Steers how patches are written. It never overrides the evidence rules - a patch is sourced and anchored first, styled second.

Writer model Which model writes the patches. It is recorded on the run, and we never quietly substitute another one.

What we check, and the evidence for it These are not our opinions. Each factor names a published study and the effect it measured - hover any one to read it, or follow the source link.

Three things we deliberately will not do

- credits Charged when the run starts, and refunded automatically if the run fails.

Running

Recent runs

Nothing yet.

Content Tool

Hi, I'm George.