Why Leolens exists
People ask an AI assistant about a business and act on the answer. That answer is assembled from whatever the engines have read, and for a business with more than one location it is frequently wrong about a specific one — the insurance a practice takes, the hours of one office, whether a hotel allows dogs. Nobody is checking. Leolens checks, quotes what was said, and writes the correction.
What we will and will not say
- A claim is wrong only when we can prove it — the engine's words have to contradict a fact you declared, quoted verbatim from the answer they appeared in.
- Unverifiable is not wrong — where a statement cannot be checked against what you told us, it is reported as unverifiable, and where you declared nothing it is reported as not measured. Neither is counted against you.
- Every figure carries its count — each question is asked three times per engine, and every number on a report shows how many answers it came from, on which engines, and when.
- No score — there is no single accuracy number. A composite would be unfalsifiable, and this product is the claim that our numbers can be checked.
- No invented data — we do not publish estimates of how often people ask about you, because that data does not exist.
Who runs it
Leolens is independent and small: the engine, the reports and the corrections are built and run in-house, not resold from someone else's data. Questions reach a person, not a queue — write to hello@tryleolens.com and you will get a straight answer, including "we cannot measure that".
Named after Leo — our Samoyed and Chief Watchdog Officer. Samoyeds were bred to keep watch, and that is the whole product: a tireless watchdog on what the machines say about your business. (Yes, he's a very good boy.)