An ACN, a thesis, and no product.

An agent finishes a task by writing a message. That message is generated by the same process that generated the work, under the same pressure to produce something that reads as complete. Treating it as a report on the work is a category error, and it is the default in most tooling because the message is cheap to read and the machine state is not.

Everything the company intends to build follows from taking that seriously. If the closing message is unreliable, the evaluation has to come from somewhere else, which means a sealed environment, a recorded transcript, a diff of what changed, and conditions written before the run rather than inferred after it.

Why claim reconciliation and not just assertions

Assertions alone would be enough if the only thing that mattered were the end state. In practice people read the summary and act on it, so a false summary attached to a technically passing run still causes damage. Counting supported claims against contradicted ones puts a number on that, and the number is different from the pass rate often enough to justify the extra machinery.

What would show the thesis is wrong

If contradicted claims turn out to be rare once runs are sealed properly, or if the contradictions are almost all trivial phrasing rather than substantive misreporting, then the extra stage is not earning its cost and should be dropped. That is a measurable outcome, and the intention is to publish it either way.

Prior art, honestly

Sandboxed task evaluation is not a new idea, and neither is diffing a container. Several open benchmarks already run agents in disposable environments and check end states. The narrow contribution being attempted here is the reconciliation stage and the habit of publishing unresolvable claims as their own figure. If that turns out to exist already in a form that works, the right response is to use it.

No fabricated proof

No invented clients, no testimonials, no user counts, no download figures, no logos of organisations that have never heard of us, and no certifications. Where this site describes something that does not exist yet, it says so in the same sentence rather than in a footnote.

No contact form

A form either posts nowhere, which is a lie in HTML, or posts somewhere and quietly collects an address and a browser fingerprint that nobody agreed to hand over. One published email address is honest and works with anything.

No analytics on this site

No measurement script, no advertising pixel, no session recording. That is also why there is no consent banner, and the reasoning is set out in the cookie notice.

No customer data yet, and a plan for when there is

An evaluation run would carry someone else's code, prompts and internal file names. The privacy policy already sets out how that material would be handled, in advance of holding any of it, so the first customer has something to read before they decide rather than afterwards.