How it works

How we ground a comparison

Every comparison is generated by an AI model, then labelled according to where each value actually came from. This page explains what those labels mean and where the limits are.

What happens when you compare

Your items and any context you describe are sent to Google's Gemini model with web search enabled. The model searches for current information about each item, then returns a verdict and an attribute table built from what it found.

Results are cached. If someone has already run the same comparison recently, you get that stored result instantly rather than a fresh generation — the content is identical, and it does not count against your daily allowance.

What the source badges mean

Every value in the attribute table carries a badge showing where it came from:

Vendor-verifiedThe value appeared in the search results on the item's own official website or documentation.

Found on webThe value appeared on a credible third-party source — a review site, reference database or news outlet — rather than the vendor's own pages.

Unverified — check sourceLive web search was unavailable for that request, so the value comes from the model's training data and nothing corroborated it. Treat these with care.

When a fact genuinely is not published anywhere, we say so rather than inventing a plausible number.

Where this falls short

Comparisons are AI-generated and can be wrong. Pricing changes, specifications get revised, and search results are not always complete. Check anything that matters against the vendor's own page before you commit money to it.

A badge records where a value was found, not whether it is still accurate today. Web search sometimes hits capacity limits, in which case the comparison still runs but its values are labelled unverified.