CiteGraph

CiteGraph shows founders and marketing teams whether AI assistants recommend their product, and what to do when they do not. It starts with the questions your buyers actually type into ChatGPT, Claude, Gemini and Perplexity: the compare, alternative-to and which-should-I-use questions in your category, mined from your own site and from public threads. CiteGraph asks each one on all four engines, several times, through the official APIs, and keeps every answer verbatim. Nothing is scraped or simulated. From those answers you get three things no score can give you. First, your blind spots: every question you lose, the exact answer the engine gave, the rivals it named instead, and the pages it built the answer from. Second, the diagnosis: why the engine picked them, the one sentence about your product that would have put you in the answer, and where that sentence needs to live, on your site or on a third-party page the engines already trust. Third, the fix itself: CiteGraph writes the page the engines would cite, drafted from the answers you lost and your own site, with every sentence traced to its source and nothing invented. Everything is measured, not guessed. Rates are sampled across repeated runs and printed with their error bars. Weekly re-scans show whether a change moved anything, with a digest only when something actually changed. A cited-source map grades every page the engines lean on by what it says about you. Share of voice tracks the rivals you choose. There is also Ask, an assistant that reads your scan and acts on the plan, and an MCP server that puts the same tools inside Claude, Cursor and VS Code, so a coding agent can apply the readiness fixes in your repo. Two free checks need no signup: a live visibility check that asks one real buyer question and shows the answer, and a readiness check with fixes. A full scan is $12 once, with no subscription. Plans start at $29 a month with a 7-day trial.