SERP data
Serper
Live SERP lookups when a check has to be current rather than cached.
Feature
See whether the AIs name you, and who they name instead
Cited by ChatGPT, or not · Runs in Claude Code
What it does
Ask ChatGPT, Gemini, Claude and Perplexity the questions your buyers actually type, and record who gets named. The system mines those prompts from your own Search Console data rather than inventing them, measures your share of the answers, breaks down the competitors cited ahead of you, and rewrites a page so it can be quoted.
$ /prompt-mining --site yoursite.com
mined 41 prompts from GSC, weighted by GA4 page value
$ /ai-visibility --site yoursite.com
cited in 9 of 41 answers · 3 competitors ahead of you
§ Scope
11 Skills involved
1 Specialist agents
4 Data sources read
§ Limits
What this does not do
Stated plainly, so you do not find out later
AI answers are not deterministic: the same prompt can return a different set of citations an hour later. Tracking does not paper over that. A run is a sample, movement between runs is only reported once it clears a significance test, and every cell in the matrix links to the answers behind it.
§ Skills
11 of them
The commands behind it
Each one is a single command against real data. They compose: the output of one is the input of the next.
Measure your share of AI answers
How often you are named across the AI engines, and who gets named instead.
You get back An AI visibility report.
Watch your AI citations move over time
Turn a one-off AI visibility check into a scheduled trend across ChatGPT, Gemini, Claude and Perplexity.
You get back A prompt-by-engine matrix, and what actually changed.
See who is buying ads inside ChatGPT
Check which brands run paid placements on a prompt, and whether each ad really appeared on screen.
You get back Advertisers per prompt, rendered or suppressed.
See what the AIs actually answer
Ask ChatGPT, Claude, Gemini and Perplexity a question and read the sources each one used.
You get back Each model's answer and its cited sources.
See who the AIs cite instead of you
Find the competitors AI engines name for your topic, and what their pages do that yours does not.
You get back Competitors, their articles, and your gaps.
Rewrite a page so AI engines can cite it
Restructure content so an answer engine can lift a passage out of it cleanly.
You get back The page, restructured for extraction.
Work out what people would ask an AI
Derive the prompts people would actually type, from your own Search Console and Analytics data.
You get back Ranked prompts, weighted by page value.
Find questions you should also rank for
Expand what you already rank for into the related questions nobody has covered.
You get back Uncovered questions that fit your site.
Measure how often AI engines mention you
Ask ChatGPT, Gemini, Claude and Perplexity buyer-intent questions and count who gets named.
You get back An AI visibility score, with a spend ceiling.
See traffic coming from AI tools
Check whether ChatGPT, Perplexity and Gemini send you visitors, and whether it is growing.
You get back Referral traffic by AI tool, over time.
Find what people really complain about
Real questions and frustrations from Reddit threads in your niche.
You get back Questions and pain points, with sources.
§ Agents
1 on it
Who does the work
An agent holds a role rather than a command: it decides which skills to call, in what order, and what the results mean together.
§ Sources
SERP data
DataForSEO
Keyword volumes, competitor rankings and SERP features.
Crawl
Firecrawl
Full-site crawls and page extraction, including rendered content.
Analytics
Google Analytics 4
Sessions, conversions and referral traffic through the Data and Admin APIs.
§ Elsewhere
The rest of what it does
Backlink Analysis Who links to you, who hurts Rankings & Search Console Your real positions, read properly Technical Site Audit What is quietly broken Keyword & Topic Research What to write next, and why Content Production Brief to published, with images Competitor Intelligence Their plan, taken apart Client Reporting The monthly deck, written