New Chat
Paste a draft. Get a measurable quality score, see why it scored that way, and know what to fix first. Separate checks flag AI-tone, evidence and publishing risks — without mixing them into your quality score. See how the check works →
ForgeRank AI doesn't reduce content quality to one vague “AI score”. Every piece is read on four named dimensions, and you can see the reasons behind the number.
Adds what the top of search doesn't already say — no restating, no padding.
Names versions, prices, dates and mechanisms instead of vague claims — details you can check.
No invented data, quotes or people. Claims ship only with the attribution they actually have.
Every claim carries a why — the mechanism, the cause and the consequence.
ForgeRank AI turns a content score into a publishing decision. The free check at the top of this page already returns the first three panels — score, diagnosis and the single highest-priority fix. Signing in adds the evidence check and keeps every score in your history.
One unedited capture of a single run on this app: the topic goes in, the draft streams out — then the same draft is scored, diagnosed and re-scored, with those numbers printed right below it. Nothing hand-assembled, no invented data.
The four failures below are the ones a quality gate actually addresses — AI-flavoured phrasing, claims you can't trust, pieces that look complete but say little, and tools that generate without ever diagnosing. For each one: what goes wrong, what ForgeRank AI does about it, and where we admit the limit. Every mechanism is real and runs on your own draft — score one free with the Analyzer at the top of this page.
The same loop runs when you start from a blank page: the draft comes out of generation, but it still has to clear the gate before it leaves. Generating is optional — you can score an existing draft, or a competitor's piece, and never generate anything. Skip to real scored output →
Paste any article or post into the Analyzer — a competitor's piece, a top-performing post, or your own draft — and get a Content Quality Score on the same 4-dimension rubric the generator uses, plus source recommendations, AI-Tone patterns and publishing-readiness checks. First scores are free, no card.
Only if you need a first draft: ForgeRank AI writes a platform-native one for Google, X, LinkedIn, YouTube, Instagram, TikTok, Facebook or Reddit — one credit per piece, no prompt engineering. It leaves through the same gate as everything else.
Drafts below their quality bar are rewritten automatically at no extra credit, and AI-Tone phrasing is scrubbed before delivery — here's a real X draft that climbed 6.2 → 7.6.
Re-run the gate and decide. Then ship with the publish pack — title options, tags, quotable lines, best time to post, sourced pull quotes — plus a Deep Insight Report that shows the why. See the full publish-pack checklist →
One brief, three platforms: the X, Instagram and TikTok pieces below all came from the same brief — "3 things that make AI content feel fake" — each rewritten for its platform, then scored by the same gate. The Google article is live on this site — open it. Every score on these cards is a recorded product run with its date; each card links to its full scored record on our receipts page.
Why the viral "toothpaste hack" is wrong — and the two-cycle method that actually protects seals, pumps and sensors.
Read the live article → (recorded editorial QC run)"AI content reeks when it's 1) allergic to periods, 2) allergic to opinions, and 3) allergic to being wrong. Real humans have edges. Mine is the missing bad take."
X · short-form rubric — Ready, lint clean → full scored record"You can spot AI content in about three seconds, and grammar isn't the real giveaway. The tells are all in what's missing."
Instagram · short-form rubric — Ready → full scored record"Nobody talks like a grammar checker. Real people say 'gonna,' trail off mid-thought, double back to fix what they meant."
TikTok · short-form rubric — Ready, lint clean → full scored recordThis is the case study, with no invented customer: the washing-machine how-to above carries editorial QC 8.2/10 — the same Analyzer you can try free, re-run on the published text (run dates on /receipts) — and it's live on this site. Read it → · our verification receipts → · score your own draft free →
x_posts/) — see them explained on the report hub →
Not just text. A scored readout on our four dimensions — real information gain, specificity density, source compliance, and argument depth — so you see exactly what a draft does well and where it falls short before you publish it. Those four dimensions and their weights are ForgeRank's own evaluation model, not Google's ranking formula. The card below is the washing-machine how-to's editorial QC score from the site's QC record (scripts/_blog_editorial_qc.json) — the same 4-dimension gate that powers the Analyzer. Every run date is listed on /receipts.
Same gate ships with every piece — Google, X, YouTube, Instagram, TikTok, Facebook, Reddit each scored on the 4-dimension rubric with its own pass line. And the Analyzer scores any article you paste with the same rubric. How we score this → · see a real scored report →
One credit = one publish-ready pack on any channel — a 280-char X post, a LinkedIn carousel, a 5,000-word Google pillar or a YouTube script. No word-count tiers, no extra charges.
Every score we publish comes from a recorded run in this repo, filed with its date and file. Nothing here is a made-up figure — and we publish the low scores too.
Browse recorded runs and QC records: real scored report → · 35 QC'd topic articles →
Built in public — this whole SaaS (quality gate included) was made in 12 days. Read the free chapters →
Your card details are never stored by us. Taxes are included where applicable. Credits never expire. Read our Refund Policy & Terms.