I'm Erik Rekola, and this is my one-person business in Tampere, Finland. I check what AI assistants can read about a product, and where a website and its API give different answers about the same thing.
I spent six years in industrial engineering and service work at UPM, Franke, Thermo Fisher Scientific and ASM International, in roles that depended on reliable machine data. Now I check the data AI agents read.
A product page can look clear to a person while its structured data or API gives an automated client a different price or stock status. An assistant can repeat the wrong one.
The audit compares those sources and records what selected AI assistants say about the product. The written report shows the evidence and the order to fix things in.
Our stack
The site is one Cloudflare Worker. Page content is written as Markdown inside the Worker, which renders it as HTML for people and returns the Markdown itself when a client asks for text/markdown. The homepage also answers with a JSON summary.
For agents the site publishes JSON-LD, llms.txt, an OpenAPI description and discovery files under /.well-known. The MCP server runs on a second Worker, and the source for both is on GitHub.
