About

Someone with real knowledge of a thing says something specific about it, out loud, in a two-hour video or a long review. That sentence is worth more than a page written to rank for it. The problem is that it is unfindable: it is forty minutes into a transcript, it is one of thousands, and nothing on the open web tells you that the person saying it has cooked professionally for fifteen years or that a brand paid for the pan in their hand.

So this site reads long-form sources and pulls out the individual claims: who said it, about what, what kind of claim it is, how strongly, why, and whether they disclosed an interest. Each one keeps the verbatim quote and a link to the second it was said. The result is a graph you can ask questions of — what do chefs reach for, what do chefs and people who keep things for twenty years agree on, what is the common objection to the thing everyone recommends.

Today that is 3418 claims from 78 named people, quoted against 4893 evidence spans in 151 sources, all of them on YouTube.

Not a review site. We test nothing. We own none of these products and have no test kitchen. Everything here is somebody else's testimony, structured.

Not an AI content farm. A language model does the reading — that is the ingestion mechanism, and the method page says exactly how — but a model is never the source of a claim. Every claim on this site traces to a named human in a citable source at a specific timestamp, and a claim with no quote behind it cannot be stored at all.

Not a scoring site. There is no Reached For number out of ten. Signals are never summed into one, because summing them decides for you which ones mattered. Durability, professional use, price, how long someone has actually owned theirs and how loudly people object are shown separately and left separate.

Not an affiliate blog. We expect to earn affiliate revenue and the disclosure page says how. A commercial link is attached to a product, never woven into a sentence, and it never changes what the evidence says.

Reached For is built and run by Zach Blender. One person, no staff, no newsroom, no commercial relationship with any brand named on the site.

Nobody here is a cookware expert, and that is the point: the expertise on this site belongs to the people quoted on it. It is also, openly, an experiment in whether AI can be pointed at something more useful than writing more articles. Every page here was assembled by software, and the step that reads a transcript and decides what was claimed is a language model. That is a thing worth being suspicious of, so the site is built so that you do not have to take its word for anything: a claim cannot exist here without a quote attached, every quote links back to the source at the timestamp it was said, and what a model produces is treated as derived data rather than as fact. The aim was never to generate opinions about pans. It was to take opinions people had already published, and make them countable, comparable and checkable.

That has one consequence worth stating plainly: the judgement in this product is in the pipeline, not in an editor. Nobody picks the winner. What gets published is decided by whether the evidence clears a threshold, and pages that do not clear it are not written thinner — they are not written.

516 pages have cleared it. 733 candidate pages have not, and have no URL.

How a claim gets in — the extraction, the evidence rule, the gate, and the things we know are wrong with this corpus today.
Money and disclosure — what we earn, how a paid voice is labelled, and why we never delete one.
Corrections — we publish factual statements about named living people. Here is how to tell us we got one wrong.