Content Creator

AIContentGapAudit

Diagnose what your page is missing for AI search, then generate publish-ready content to fill the gaps.

Beta v0.9 - Single Keyword Analysis

FAQ

FAQ&Methodology

The Shift

SEO is being re-indexed by AI assistants

58% of Google searches now end without a click. When your customer asks ChatGPT a question, your content either gets cited or it doesn't. The rules are different from classical SEO, and most content tools haven't caught up.

01

Reads well, doesn't get cited

AI assistants pull from passages with atomic claims, specific examples, and clean schema. Long generic paragraphs lose every time.

02

Generic AI output gives you away

"Leverage", "ever-evolving", "crucial", em dashes. These tells mark text as AI-generated and lower its citation rate.

03

Auditing is decoupled from fixing

Most tools tell you what's broken. They don't write the fix in the same workflow. You audit, then go open another tool.

04

Schema is an afterthought

FAQPage, BlogPosting, and ItemList schema decide whether AI can parse your content. Few writers add them automatically.

Five modes, one tool

Every step of the audit-to-publish loop

Diagnose finds the gaps. Four generation modes fix them. Each runs the same 5-pass pipeline that strips AI fingerprints and bakes in schema.

01
Diagnose
Public · 3 free per IP per day
Audit any URL against a target search phrase. Returns specific gaps that prevent AI citation: missing schema, weak comparisons, absent specifics. Each gap has an estimated impact.
1 crper diagnosis
02
New Blog Article
From diagnosed gap
Generates a full article from a diagnosed gap. Includes H2/H3 structure, FAQ section (optional), FAQPage + BlogPosting schema, internal links, and meta tags. 1200-2400 words.
4 crper article
03
Direct Generation
From topic, no URL needed
Write a publish-ready article on any topic. No URL, no diagnosis. Same 5-pass pipeline. Best for when you already know the topic but don't have a comparison page to audit.
4 crper article
04
Rewrite
Preserves your facts
Rewrites an existing article for AI readability. Pro draft preserves every specific fact from the original. Choose tighten (cut 20%), match (same length), or expand (add 20%).
4 crper rewrite
05
On-Page Section
WordPress, Shopify, WooCommerce, Wix, Webflow
Landing-page or product-page section, formatted for your platform. Outputs platform-native markup (Gutenberg blocks for WordPress, Liquid-compatible for Shopify) plus matching schema.
4 crper section

Methodology

The four properties that decide whether AI cites you

GIST is the scoring framework behind every output. Each generated passage is tested against these four properties before it ships.

G

Granularity

Each claim is atomic and self-contained. AI assistants pull single sentences or short passages, not paragraphs.

Weak: There are many factors...
Strong: Page speed under 2.5s improves rankings by 8-12%.
I

Inferability

Surrounding context is enough to verify the claim. The passage stands alone when extracted.

Weak: As mentioned above, this tool is faster.
Strong: Screaming Frog crawls 500 URLs per second on a 16GB machine.
S

Specificity

Concrete tool names, numbers, file formats, version numbers. No generic abstractions.

Weak: Use an SEO tool to find broken links.
Strong: Use Screaming Frog or Ahrefs Site Audit to find 4xx errors.
T

Topicality

The section answers the query it sits under. No padding, no off-topic context-setting paragraphs.

Weak: How to fix redirect chains → 2 paragraphs about why SEO matters.
Strong: First sentence is the fix.

Pipeline

Five passes per article, different models for different jobs

A single LLM call cannot produce publish-ready content. The pipeline routes each job to the right model: fast for analysis, capable for writing, deterministic for cleanup.

01
Strategy
Flash

Identifies angle, voice, and which concrete examples to weave in.

02
Draft
Pro

Writes the article. The most capable model handles the hardest job.

03
Specificity
Flash

Replaces abstractions with concrete tool names, numbers, examples.

04
Strip Tics
Flash + Regex

LLM removes tics. Deterministic regex sweep removes the rest.

05
Final Pass
Flash

Smooths transitions. Generates meta tags, slug, schema.

FAQ

Questions worth answering

What does GIST stand for and what does it measure?+

GIST stands for Granularity, Inferability, Specificity, and Topicality. These four properties determine whether a passage gets cited by AI assistants like ChatGPT, Perplexity, and Google's AI Overviews. Pages that score high on all four get cited; pages that don't, do not.

How is this different from generic AI writing tools?+

Generic AI writers produce text. This tool produces text engineered to be cited by AI assistants. The pipeline runs five passes per article: a strategy pass to choose the angle, a draft pass on the most capable model to write it, a specificity pass to add concrete tool names and numbers, a fingerprint strip to remove AI tics, and a final coherence pass with schema and meta. Every output passes a deterministic banned-word sweep to remove phrases like "crucial", "leverage", "ever-evolving", and em dashes that mark text as AI-generated.

What modes does the tool support?+

Five modes. Diagnose audits any page against a target phrase and returns specific gaps. New Blog Article generates a full article from a diagnosed gap. Direct Generation writes an article on any topic without a URL or diagnosis. Article Rewriter rewrites an existing article for AI readability while preserving facts. On-Page Content generates landing or product page sections, optionally formatted for WordPress, Shopify, WooCommerce, Wix, or Webflow.

How does pricing work?+

The Content Gap Audit is included in the Website AI Score subscription at $49 per month for 100 credits. A diagnosis costs 1 credit. Any generation mode costs 4 credits. A full audit-to-publish workflow on a single page is 5 credits. Unused credits do not roll over.

Will the output need editing before publishing?+

Most outputs are publish-ready. The pipeline produces 1200 to 2400 word articles with H2 and H3 structure, real tool names rather than generic terms, valid schema markup, meta tags, and FAQ sections on request. Author and publisher names can be baked into the schema during generation.

What if the worker fails or generates poor output?+

Credits refund automatically on any API failure, parse error, scraper block, language drift, or output that comes back too short to be useful. The refund is binary: either the output is delivered or the credit is returned. There is no partial-quality logic.

Can I rewrite content I don't own?+

The rewriter accepts any public URL the scraper can read. Use it on your own content, content you have rights to, or competitor content for analysis. The tool doesn't enforce ownership; you do. Republishing someone else's content as your own is a copyright issue regardless of how it was processed.

Does it work for languages other than English?+

English only for now. The pipeline includes a language detection step that returns a refund if the input or output drifts away from English. Multi-language support is on the roadmap; it requires tuning the banned-phrase sweep and tone presets per language.

Stop guessing. Run a diagnosis in 30 seconds.

Free check at the top of this page. If the gaps look real, the fix is one click away inside the full tool.