Who is it for?
- Founders sizing up a market before building
- Product managers deciding what to fix or build next
- Marketers looking for a real, evidence-backed point of difference
- Researchers and analysts doing competitive work
- SaaS companies and e-commerce businesses evaluating competitors
- Agencies doing competitive research for clients
What does it analyze?
Any page with customer reviews — a Trustpilot profile, an app store listing, an Amazon product page, a G2 or Capterra comparison — up to several hundred reviews per run. Each review is individually classified by category, sentiment, explicit requests, implied needs, and severity.
What does it produce?
A structured report: an overall sentiment/risk read, a pain map of categories by frequency and severity, ranked opportunities with the evidence behind each one, real customer quotes, and a one-click PDF export. Nothing in the report is presented without the count or quote that backs it.
How does it differ from asking a general AI to summarize reviews?
A general-purpose chatbot given a pile of review text will summarize it — but it's also doing the counting itself, in its head, which is exactly where language models are least reliable. Ask it "what percent of these are negative" on five hundred reviews and it's estimating, not counting.
Review X-Ray separates the two jobs. A fast, typed decision model classifies every review individually — this is counting, not writing. Only after those classifications are tallied in plain code does a writer model turn the resulting numbers into prose. It's never asked to count anything itself, and it's explicitly told not to invent a number or a quote that wasn't handed to it.
Review X-Ray is a review-analysis tool built around the discipline that counting and writing are two different jobs — one for a classifier, one for a language model — so the report stays traceable back to real reviews, not estimated from them.