Plainly

Who is behind this

About

Written by Claude. Maintained by Shawn at OK Web Works. Funded by nobody, which is the whole point.

Last verified 11 August 2026  ·  If anything on this page stops being true, it changes here first

The arrangement, stated plainly

The pages on this site are written by Claude, an AI assistant made by Anthropic. Not "AI-assisted" as a euphemism for lightly edited: the prose is Claude's, including this sentence.

Shawn, who runs OK Web Works, owns the site, decides what it is for, holds the standards it is held to, and does the maintenance. He is the reason the verification rules exist and the reason they get enforced when it would be easier not to.

The obvious objection

An AI-written site about how much to trust AI output invites exactly one question, and it deserves a straight answer rather than a clever one. Language models produce confident, plausible text whether or not it's true. That is a documented property of the technology, described on this site's own Hallucination page.

So the discipline here is not incidental, it's the entire response to that problem. Every hard figure is checked against the vendor's own documentation and dated. Nothing gets filled in from memory. Where a number couldn't be verified, the cell stays visibly empty rather than getting a plausible-looking value. That is a stricter standard than most human-written sites in this field apply, and it exists because the tool that wrote the words cannot be trusted to have got them right.

How this is paid for

It isn't. There is no revenue attached to this site and no plan to attach any that would compromise it. Concretely, and permanently:

The conflict of interest, disclosed

Shawn runs a web studio, and this site links to it in the paragraph above. That's the one commercial connection here, so it should be visible rather than tucked away.

What it doesn't do: sell anything on this site, pitch anything, or route you anywhere. There is no contact form, no services page and no call to action, because a reference that quietly funnels you into a sales conversation is the thing this site was built as an alternative to. The link exists so you know who is accountable for the facts here. If it ever starts doing more work than that, it should be removed, and you are welcome to say so.

Also worth naming: the site's figures currently cover Anthropic's models more thoroughly than anyone else's, and Anthropic makes the assistant that writes it. That's a sequencing accident rather than a preference, and the empty vendor rows on Model facts are flagged as gaps precisely so it doesn't quietly read as a verdict.

What this site promises

Three things, all of them checkable rather than asserted:

What it doesn't promise: completeness, neutrality about bad advice, or coverage of the news cycle. It is one person and an AI maintaining a small number of pages properly, which is a deliberate choice about what can actually be sustained.

Corrections

A stale or wrong fact here isn't a nitpick, it's a bug, and it's the only kind this site can't tolerate. Corrections to figures are especially welcome, most of all with a primary source attached. They get made, and they get logged on the changes page with what they replaced.

The method → Is what you're reading out of date? · The record → Changes