It doesn’t just write. It watches what works and decides what to write next.
Rank Tuner mines keyword gaps, researches each topic against live sources, drafts in your team’s voice, routes every piece through a human gate, publishes on schedule, then reads its own results to re-score what comes next. A closed loop, not a one-shot generator.
Five jobs, one continuous system
Each stage feeds the next. The system runs the work; a human holds the gate; the loop closes when measurement re-scores what to write. None of it publishes on its own.
Mine the gaps
Rank Tuner pulls competitor keyword gaps and the queries you already get impressions for but no clicks, then scores and clusters them into a backlog of briefs. It starts from where the opportunity already is, not a blank page.
“It starts where you already almost rank.”
Research against live sources
For each topic the model researches the live web with citations, capping searches and preferring trusted domains. Every source is logged to the task’s provenance row, so the draft is traceable to what informed it.
“Grounded in sources, not guesses.”
Draft in brand voice
Drafting uses a persona built from real published work plus a voice spec, with the static context prompt-cached so each article stays cheap and on-voice. Output is structured Portable Text, ready for the CMS.
“Your voice — learned, not approximated.”
Publish on schedule — verified
Approved articles publish to the CMS on a schedule, then the system re-fetches the live URL and confirms the revision, canonical, and a 200 before it marks anything published. It is verified, not fire-and-forget.
“Published means checked, then confirmed.”
Track rankings, traffic & AI citations
A daily pull aggregates rankings, traffic, and AI-answer citations onto one timeline, with publish events and Google core-update markers overlaid. Gaps render as gaps, never as zeros.
“Every publish, measured against the market.”
A person signs off on every word. That’s the point.
A claim-validation pass extracts every factual and client claim and grounds it — allow-list match for client claims, web search for facts. It only ever flags; it never auto-approves. A human approves, edits, or rejects, and nothing reaches the CMS without that sign-off. Search engines penalize thin content produced at scale, not careful work — so a human-approved, attributed, low-cadence pipeline is the defensible posture, not a limitation.
How the gate worksThe brain reads its own results and steers
A daily job reads rankings, traffic, and AI citations against every publish event and core update, then re-scores the backlog: double down on winners, refresh decaying pieces, drop cannibalizing topics. Decay is detected statistically — period-over-period, slope, CTR-vs-impression divergence — never a hardcoded threshold. The output of measuring becomes the input to mining. That is the loop.
See the loop close“I approve every word before it goes live, and the system tells me what to write next.”