Turn one gap into a page you can stand behind — then keep it current

A workspace for the questions your buyers ask where AI answers leave you out, for a Google Search Console demand gap, or for a topic you bring yourself. Each piece starts from evidence you can inspect, is written against numbered sources, and cannot be released while a fact-check reports unsupported claims unless a person records an override. After publishing, the page is tracked at its live URL, and a page that goes stale is refreshed in place rather than re-created. Whether an engine ends up citing it is not ours to promise — everything you control before and after release is here.

Real AI queries · no credit card · 3 free checks a day

What you get

Draft, edit, publish — in that order

Create draft, Edit draft, Publish: each is its own view of the same record, and the stepper shows where you are and what is still locked. Publish is unavailable until the piece has been edited, so a rough draft cannot leave by accident.

A queue that names the stage of every piece

The workspace sorts your work by what it needs from you: needs attention, published and waiting for a citation, published but not cited yet, and cited. Each row carries its live URL and either its published date with a staleness warning or the days since a citation was recorded, so the next page to touch is the top of a list rather than a guess.

Gaps from your monitoring, from Google, or from you

Opportunities come from the queries where your brand is missing or outranked across the 8 engines you monitor (four gap tiers), from Google Search Console demand gaps, and from pages AI already cites for your category — not from a generic keyword list. It is not gaps-only either: New article starts from any topic you bring.

A suggestion has to show its evidence

Each suggested topic states what was observed: which platforms saw the cited page and on which dates, how often, which domains AI cited instead, and the source URL it came from — plus confidence with its components, or a plain sentence saying why no score is offered. A suggestion is never a black box.

Drafts written against numbered sources

A draft is built from the material it is given — your uploaded documents and live research read as full page text — with each claim pointing at the numbered source it came from, so any sentence can be followed back to its basis. The markers are a working aid and are removed on the Publish view, which shows the page exactly as it will go out.

Your own documents, in the draft

Upload the pricing sheets, product docs, brand guides and case studies a piece should stand on, and a draft is grounded in the passages that match what it is writing about rather than in the cover of a file. Every library source links back to the document and the section it came from, and the library states how much of a document can reach a draft.

A brand voice you can write down

A style guide, an audience, do and do-not rules, and a few passages of your own prose. When a brand voice exists it decides the tone; a persona preset is only a seed for accounts that have not written one. The house style that removes the usual AI tells is on by default and can be switched off if your brand uses those constructions deliberately.

You keep or undo every change

The article is the interface: a rich-text editor with a keep-or-undo decision per proposed change, a toggle between inline marks and a clean read, and your grounding sources plus internal-link suggestions in a side drawer — so reference material never crowds the draft.

Nothing ships with a claim it cannot back

Before release the finished article is audited against its sources, and every statistic, ranking, date and unsupported claim is listed. Publishing stays blocked while one remains. If you judge a flag wrong you can override on publish, and the override is written down.

Published is a stage, not the end

Send the piece to your own site or over a webhook, and have it picked up with IndexNow. Every query then carries one coverage state — not started, draft, published and awaiting citation, cited, or stale — so a covered gap is never handed back as “write another article”. A page still uncited after 30 days becomes stale: Refresh regenerates the draft in place, you edit and re-check it, and publishing the update replaces the live page at the same URL with the previous draft kept as a revision. Refreshing several stale pages regenerates their drafts without publishing any of them.

See it in the product

Review suggestions dialog with 26 topics to approve. The open Content opportunities group shows a card for “Best AI visibility tracking and answer engine optimization tools”, observed 5× on Microsoft Copilot and Perplexity between 31 Aug 2026 and 22 Oct 2026, citing aeospotlight.com (37×), youtube.com (27×) and tryfound.com (12×) instead, sourced from rankability.com, with evidence 67%, gap severity 30% and competitors cited 100%. — You approve a topic because you can read why it was suggested — which platforms and dates produced it, which pages AI cited instead, and how strong the evidence is.

The “Optimize — published but not cited yet” band holding 4 stale pages: AEO Spotlight vs Brand24 for AI visibility monitoring (51d without a citation), the best AEO tools guide (54d), a citation-signal piece (47d) and a new-domain authority guide (31d), each marked Stale with a link to its live article. — A page that stops earning citations keeps its live URL and shows its days without one, so refreshing is a deliberate edit of a real page — never a silent re-write.

Edit step of the draft “Is AI Visibility Worth It? A Practical Guide for 2026”, last updated October 21, 2026: the stepper shows Create draft done and Edit draft current with Publish not yet available, and the Suggested changes panel reports 5 of 5 changes waiting with Accept all and Reject all above the article text. — The draft is the interface: proposed edits sit in front of you as keep-or-undo decisions, and Publish stays locked until the piece has been edited and checked.

How it works

Create the draft

Pick what you are aiming at — a missed AI answer, a Google Search Console demand gap, a page AI cites for a rival — or start from a topic you bring. Choose your grounding (your own sources, or acknowledged live research) and the writing voice, and the platform assembles the brief and writes an answer-first draft.

Edit and verify

Work the draft in place: keep or undo each proposed change, open the sources and internal-link suggestions, then run Verify to audit the claims against the grounding. Fix or remove whatever it flags — or override deliberately, which is recorded — before anything ships.

Publish and track

Choose the destination and the byline, publish, and follow the piece from draft to cited. A page still uncited after 30 days becomes stale and offers one action: refresh the draft in place and publish the update to the same URL. Cited pages leave the actionable queue.

Why you can trust the data

Every generated claim is tied to a numbered grounding source, so a draft can be audited line by line.

A library-sourced claim links back to the document and the exact section it came from, and the library shows how much of each document can reach a draft.

Publishing is refused while the fact-check reports unsupported claims, unless a person overrides on publish — and that override is written to the audit trail.

Coverage is tracked per query, so a gap that is already covered is not offered back as “write another article”.

Your live page changes only when you publish the update: refreshing one stale page, or many, produces drafts first.

Questions

Do you guarantee the content will be cited?

No. Whether an AI engine cites a page depends on third-party models, their sources and their policies, none of which we control. What we do is show you the gaps, ground the draft in sources you can open, block release while unsupported claims remain, publish to a URL you own, and then show you whether that URL gets cited.

Does it publish without my approval?

Not by default. Autonomy starts at manual, and in propose mode the platform prepares the action and waits for you. Hands-off publishing is opt-in.

What happens when the fact-check finds a problem?

The report names the claims it could not support, and publishing stays blocked until you fix or remove them. If you judge a flag wrong you can override on publish — the override is recorded, not hidden.

Can I write about a topic that is not a detected gap?

Yes. Detected gaps are a starting point, not a restriction: New article starts from any topic, and Google Search Console demand gaps and pages AI cites for rivals become opportunities too. The same grounding, fact-check and publishing path applies either way.

What happens to a published page that stops being cited?

After a 30-day observation window it is marked stale, and Refresh regenerates the draft to bring facts, dates, schema and authority signals up to date while the topic and URL stay the same. The live page is untouched until you edit the refreshed draft and publish it, which replaces the page at the same URL and returns it to monitoring. If you would rather leave the original alone, you can deliberately publish a variant at a new URL.

Why can the suggestions list sometimes be empty?

Because the list only holds topics with real evidence: a page AI cites across platforms or days, or Google demand above a threshold. When it is empty, “Why is nothing here?” tells you what the engine had to work with — whether monitoring scope is missing, sources were only one-off snapshots, no page is cited against your prompts yet, or the demand channel is not connected.

Does it read my whole content library?

No. Documents are split into sections, and a draft is grounded in the sections that are relevant to the piece, inside a character budget — at most two sections per document by default. The library states how many characters of each document can reach a model, so that figure is never a claim that the whole file was read, and a draft is never described as knowing your brand beyond the voice you have written down.