One person, measuring in public.
Artifact Machine measures whether a machine sent to answer a question about a company can actually read that company’s website. Everything here is downstream of that one question.
What this is
A scanner, and the things built on top of it. The scanner reads a page in a real browser and scores it — how ready it is for search, how ready it is for answer engines, and the distance between the two. Everything else is a different way of getting at that number: a free check for one site, an open API for a list, a published study of 57 companies, and paid columns and audits on Apify.
What it is not
Not a platform, not a dashboard, not a monitoring service. Nothing here needs an account with us, stores your data, or emails you. If a tool would need to run continuously on your behalf to be useful, it does not get built — a service that quietly stops working is worse than no service, and we would rather ship the thing that cannot fail silently.
It also will not tell you whether an AI assistant mentions you today. That requires repeatedly sampling the models themselves, several companies sell it, and this is not it.
Pricing
The single-site check and the batch API are free and need no key. There are two paid lines, both billed per result through Apify, with no subscription and nothing to cancel:
- Enrichment columns, $0.001 to $0.003 a row. Built for a list. Priced so that scoring ten thousand domains is a rounding error.
- Single-purpose audits, $0.01 to $0.02 a result. Built for one site at a time, in more depth than a column can carry.
A run that returns no results costs nothing.
Honest status
Nothing here has earned money. The audit line has been published since July 2026 and has no paying usage; the enrichment line is newer and has none either. There are no testimonials on this site because there are no customers to quote. When that changes, this paragraph changes — and if a line is retired for lack of interest, it goes on the lifecycle page saying exactly that, in those words.
What the measurements will not tell you
Automated checks find machine-checkable problems. They do not certify accessibility, they do not determine legal compliance, and they do not replace a person looking at the page. Each tool page states its own limits, because a tool sold past its limits is worse than no tool.
The headline signal has a published limitation too. Across 3,000 randomly sampled domains the search-to-answer gap correlates 0.84 with search readiness alone. It ranks a list differently — about half the top 200 changes — but it is not an independent signal, and the study says so rather than leaving it out.
Getting in touch
There is no mailing list and no contact form, because there is nothing to put you on a list for. For anything about a specific tool, the issues tab on its Apify page reaches the person who built it, and it is the only channel that is actually monitored.