See where every sentence came from.
Inciteful harvests the client’s own live pages and cites them by default, then adds third-party authorities it fetched, classified live or dead, and read the real publisher and author from.
Content Engine
Content AI can cite, because nothing is invented.
Inciteful retrieves real sources, writes only from the facts it bound to them, and verifies every citation before anything goes live. The publish button refuses an unsourced claim or a dead link — that is the product, not a setting you switch on.
7-day trial, no card.
Content engine
From the gap to the published page.
The questions you are losing become briefs, the briefs become answer-shaped pages, and the pages publish into the client’s own stack.
Live component. Open a topic for the questions behind its rate; the flagged row is a measured gap, scored straight into a build.
Get your free report
The problem
Fluent is not the same as citable.
Three failures, all of them normal, all of them expensive.
Ungrounded pages give an assistant nothing to quote.
A model writing from memory produces confident prose with no source behind a single line. Assistants cite evidence, and there is none to find.
One invented statistic outlives the campaign.
A fabricated number in a client’s published page is a trust problem you discover months later, in a meeting, in front of the client.
Your CMS never sees the schema.
The structured data that tells a crawler what a page is usually stops at the export. Crawlers do not run JavaScript: if it is not in the HTML, it is not read.
The difference
Ask for a page on brand strategy for scale-ups.
Both sides produce a page in about a minute. Only one of them can show you where every sentence came from — and only one of them will stop you publishing when it cannot.
A general AI writer
- No source bound to any claim in it
- Invents an author and a publisher when pressed for attribution
- Nothing standing between the draft and your client’s live site
It reads well. That is the whole of what it guarantees.
Inciteful, grounded
- Reads the client’s own live pages first, then third-party sources it fetched and classified — steered to official ones on health, legal and money topics.
- Writes only from the facts it bound. The model edits fetched evidence; it is never the source of a fact.
- Re-checks every citation resolves, and refuses the publish if one does not.
It reads well too. It can also be defended, line by line.
Unsourced claim
“Rebranding lifts revenue by 23% on average.” No fetched source supports this figure.
Confirm to use — records the client company as the source
Dead citation
Citation 4 of 11 returns 404. Verified at publish time, not at draft time.
Fix the link to publish — a 404 cannot be confirmed awayCite it or cut it. There is no override, and the same gate runs on the app, on WordPress and on Shopify.
What you get
Evidence, structure, schema, gate.
Four things a client’s page needs before an assistant will repeat it.
The publish button refuses.
Placeholder or unsourced content returns a hard block instead of a warning, a dead citation fails the same way, and star ratings are emitted into schema only from real review data.
See the score before your client does.
Every page is typed — best-of, how-to, cost, compare, problem — and scored for the structure an assistant can actually lift: the answer first, headings that match the question, a real FAQ.
Ship the markup, not a ticket.
Article, FAQPage, HowTo, Product and BreadcrumbList JSON-LD can ship with a web page when its approved evidence supports that type. Writing Product/Collection metafields onto exact Shopify catalog resources belongs to the separate AI Shopping workflow.
How it works
Citable pages, built from evidence.
The order matters. Every step below happens before a word is written, or after it and before anything is published.
-
Scan the site
Inciteful reads the client’s real pages, pulls verifiable facts with a proof URL against each one, and suggests the topics worth owning.
-
Ground the facts
It fetches sources, classifies each as live, dead or unverified, and reads the genuine publisher and author off the page rather than inventing one.
-
Generate the cluster
A coherent set of typed pages with internal links, meta, FAQ and JSON-LD — up to 12 pages a cluster on Solo, 24 on Pro and Agency.
-
Gate, then publish
Citations are re-verified and every numeric claim checked against its source. Only then can it export, push to WordPress, or go live.
Delivery
Publish where your clients already live.
WordPress
Push a cluster to a connected site; pages arrive as Gutenberg blocks with the JSON-LD in the head.
Shopify
Publish the approved cluster as generic Shopify Pages with exact JSON-LD and independently verified storefront readback.
Exports
Markdown, HTML, JSON-LD, WordPress WXR or a zip — unbranded on every plan, yours to edit.
API, CLI and MCP
The v1 REST API, a one-file command line, and a full MCP server so an agent in the client’s repo can do it.
Separate product: Product Q&A, Collection guides, and Product/Collection metafield optimization stay in AI Shopping. A Content v2 Shopify Page release never mutates a catalog resource.
FAQ
What people ask about generated content.
Is this AI-written content?
Yes, and the page says so plainly. What differs is what the model is permitted to do: it edits facts that were fetched and bound to a source, and it is never the origin of one. In grounded mode, inventing a fact is not a switch you can flick — it is a step that does not exist in the pipeline.
Which models does it run on?
Haiku, Sonnet and Opus. Sonnet is the default on paid plans. Opus runs only on your own Anthropic key, because on ours a month of it can cost more than the plan does — we would rather say that than quietly cap you.
Who owns what it produces?
You do. Export it, push it into your own CMS, hand it to the client, or delete it. Your account export takes everything with it, and there is no lock-in beyond the month you are paying for.
Will Google penalise it?
The pages are grounded, sourced, typed and gated, which is the opposite of the mass-produced filler the guidelines are aimed at. Nobody can promise you a ranking, and we will not: this improves the likelihood that a page can be cited. It does not promise an outcome, and any tool telling you otherwise is guessing.
How much can I generate?
Fifteen clusters a month on Solo, forty-five on Pro and a hundred on Agency, at up to 12 pages a cluster on Solo and 24 on Pro and Agency. The current numbers always live on pricing.
My client is not on WordPress or Shopify. Does it still work?
Yes. Every cluster exports as Markdown, HTML, JSON-LD or WXR, the v1 API and CLI drive it from whatever stack you already run, and the MCP server lets an agent do it for you. The pages belong on the client’s domain either way — we do not host client content on ours.
Ready to publish something AI can quote?
Start a 7-day trial, no card. Generate a cluster, try to publish an unsourced claim, and watch the gate refuse it.