AI content writer that drafts from your own material and checks every claim

XeoRank's AI content writer, the Writing Agent, drafts only from your own material, such as interview answers and notes, and reviews every draft for unsourced claims, filler and placeholder text before it can be published. It is not a one-click article generator: give it a keyword and nothing else, and it will decline to write.

That is a deliberate choice. A model given only a keyword has little to add beyond what already ranks, and it is easy for it to fill the space with claims nobody can check. The Writing Agent is built around the opposite idea: the substance comes from you, and the software's job is to shape it and catch what should not ship. It is available on Pro and above.

Who it is for

• Founders and subject experts who know their topic but do not have time to turn notes into a finished page.
• Content teams who want every draft checked for unsourced numbers and filler before an editor sees it.
• Agencies writing for clients who need each piece to sound like the client and to hold up to a fact check.
• People already drafting with Claude, ChatGPT or Cursor who want a review step their assistant cannot talk its way past.

What it does

The Writing Agent has four parts. You can use any of them on their own.

Writing review

Point it at any text or public URL and it returns a slop index from 0 to 100, where lower is better, with findings tied to the exact passage. It flags:

• numbers, percentages and "studies show" claims with no source in their section;
• filler phrases and paragraphs with nothing specific in them;
• superlatives with no basis;
• placeholder text left in the copy;
• repeated sentences and repeated openers;
• question headings whose section does not answer the question.

The review measures the writing, never who wrote it. It is not an AI detector and does not label text as machine-written.

Fix suggestions

For findings you want help with, the agent proposes minimal patches: a specific find-and-replace for the sentence at fault, not a rewrite of the whole piece. Patches are never applied on their own. You or your assistant read them and apply the ones you agree with.

Drafting from your material

Inside a content brief, the agent drafts the piece from what you supply:

1. Your interview answers. The content workflow asks the author a short set of questions, such as what they expected and what actually happened, what number or artefact proves it, what they got wrong at first, and who exactly the piece is for.
2. Your rough draft or notes, in whatever state they are in.

It writes within the brief's evidence and your voice profile, then reviews its own draft and revises it, up to four passes, and reports any blockers left. If the brief has no interview answers and no draft from you, it stops and asks for them rather than inventing the content.

Voice profile

Give it 2,000 or more words of your own unedited writing, from URLs or pasted text, and it records how you write: sentence rhythm, how often you use "I", "we" and "you", contractions, punctuation habits and the like. Drafts and reviews then report how far a piece sits from your usual range, so a draft that reads like nobody in particular gets caught.

Why unsourced claims are a blocker

A claim with a source is one a reader, an editor or an AI engine can check. There is evidence it helps visibility too: in the GEO study by Aggarwal et al., adding quotations, statistics and cited sources raised a page's visibility in generative answers by up to about 40% on their benchmark, while keyword stuffing did little. Google's guidance on AI features in Search says the usual SEO best practices still apply to AI Overviews and AI Mode.

So in XeoRank, an unsourced statistic or a leftover placeholder is a writing blocker. The rankability score for a brief cannot reach ready while one remains, and publishing to WordPress stops and lists them until they are fixed or a person overrides them with a recorded reason.

How it fits the content workflow

1. Pick a topic from content opportunities, built from your Search Console data, content gaps, keyword clusters and AI answers.
2. Open a brief. It collects related queries, your own pages to link from, and the topics rival pages cover.
3. Answer the interview questions or paste your notes.
4. Draft. Your own AI assistant can write it through the XeoRank MCP server, or the Writing Agent can draft from your material.
5. Review, apply the fixes you agree with, and score the draft.
6. Publish to WordPress as a draft, then read it and publish it yourself.

What it does not do

• It does not write from a keyword alone. No material from you means no draft.
• It does not take its own figures on trust. Every figure in a draft is checked against your material and the brief, and any that is not there is sent back for revision, up to four passes. Where a claim needs a source, it flags it for you to supply one.
• It does not apply changes on its own. Fix suggestions are patches you choose to accept.
• It does not publish new pieces live. WordPress publishing creates a draft for you to review; an optimisation of an existing page updates that page.
• It does not detect AI writing, and it is not a tool for getting past AI detectors.
• It is not on Starter. The Writing Agent is on Pro and above.

Try it from a chat

"Review the draft on our /blog/onboarding-checklist page with the Writing Agent. List every unsourced claim and every filler paragraph, propose fixes as patches, and do not apply anything until I say which ones to keep."

Pricing

The Writing Agent is on Pro and above. Pro is $49 a month and Growth is $109 a month. Starter, at $19 a month for one site with a 14-day free trial and no card to start, includes content briefs and the rankability score.

See pricing for details.