The pipeline
Nine stages. You are needed for one.
Everything between a keyword and a published page, listed in the order it runs. The stage that needs you takes about three minutes and is the reason the output does not read like everyone else’s.
- 01
Brief
Your product, audiences and their pain points, competitors, market and tone.
- 02
Research
Keyword and SERP data, competitor gaps, and the statistics the piece is allowed to cite.
- 03
Clarifying questionsYou
A short list of questions only you can answer, each with a stated rationale so it does not read as arbitrary.
- 04
Outline
Title, angle, slug, meta description, primary and secondary keywords, and the section structure.
- 05
Draft
Sections are written in order, each aware of what has already been covered but not of how it was phrased.
- 06
Editorial pass
The only stage that sees the whole assembled article the way a reader will.
- 07
Validation
Word count, internal link targets, keyword placement, and any statistic that did not come from research.
- 08
Structured data
Article, BreadcrumbList, and FAQPage or HowTo where the content genuinely supports them.
- 09
Publish
To WordPress, Webflow, Shopify, Ghost, Wix, Blogger, a webhook or Zapier — on a cadence that ramps against how old your site is, on the weekdays you allow.
What's different
Built to rank, not just to publish.
Structured data on every article
Article, FAQ, HowTo and Breadcrumb JSON-LD are generated inside the pipeline and stored with the piece, so they travel to whatever you publish to.
Real named authors
Every article is bylined to a named person with a bio and a profile page — not the brand as a schema.org Person, which Google advises against.
Your first-hand experience
Before drafting, you answer a few questions only you can, and the answers go into the piece as material a model could never have inferred.
An MCP server for the loop
Claude, Codex or your own agent reads the calendar, answers questions, queues generation and controls autopilot — over the same API the dashboard uses.
Pace matched to site age
Publishing cadence ramps against when your site was first seen, so a young domain never ships at a volume that reads as scaled content abuse.
Search Console improvements
Connect GSC and it finds near-miss rankings, CTR gaps and thin pages, then rewrites them as accept-or-reject hunks — on posts we didn't write, too.
Autopilot modes
Autonomy is a per-project setting, not a personality the product has. Most projects should sit in draft-first for the first quarter and move when the output stops surprising you.
- 01OffNothing runs. The calendar sits where you left it.
- 02Approve eachNothing generates without a click. The slowest and the safest.
- 03Draft firstGenerates freely on schedule, a human publishes. Where most projects should sit.
- 04Full autoGenerates and publishes unattended. Available, and we would rather you earned your way to it.
WordPress, Webflow, Shopify, Ghost, Wix, Blogger, a generic webhook and Zapier. Each destination has its own page documenting the connection steps, the field mapping, and what that platform makes harder. See all destinations.
Reasonable objections
Does Google penalize AI-generated content?
AI-generated content is not against Google's guidelines. Content produced primarily to manipulate rankings is, regardless of who or what wrote it. In practice the risk tracks two things: publishing volume that outruns your site's history, and articles that add nothing that is not already on the first page. Cadence ramping addresses the first and the clarifying-question step addresses the second.
Do I have to answer the clarifying questions?
No. Set a project to full auto and generation proceeds without them. You lose the first-hand material that makes the piece worth reading, which is a real trade — it is set per project so you can make it differently for a flagship site and a side project.
Can I edit an article before it publishes?
Yes. In approve-each and draft-first modes an article waits in a ready state for review, with its validation warnings attached. Edits are yours; the pipeline does not overwrite them on a later run.
What happens if my publishing destination rejects a post?
The publish step fails loudly rather than silently, the article stays in an approved state, and the failure is recorded against the job. Every pipeline stage runs inside a memoized step, so a retry replays from recorded output instead of paying to regenerate research, outline and draft.
Does it write in languages other than English?
Language and market are set on the brief per project, and research runs against the location code you choose. Quality varies by language the way it does for any model-backed tool, so check the first few pieces in your language before scheduling a quarter of them.
Can it match how we already write?
Supply writing samples and the pipeline distills a style profile once — register, person, sentence-length distribution, habits to keep and things to avoid — then reuses it as a cached prompt prefix. Voice matching therefore costs effectively nothing per article after the first one.
Thirty credits, no card
Three articles is enough to judge it.
Sign up, give it a brief, answer a handful of questions and publish. If the result is not better than what you are running now, you have lost an afternoon.
3
Free articles on signup
- No credit card
- Published to your own site
- JSON-LD and a named author
- Cancel anytime