Built so local businesses are usable by agents, not just humans

AI agents are starting to do the booking. “Find me a plumber who can come tonight and book it.” “Is the hardware store open and do they have a 15mm elbow — reserve one.” The agent goes out, reads the web, and tries to act on the customer’s behalf.

The problem for a local business: agents can read your marketing, but they can’t tell what you’ll actually let them do. So they move on to whoever made it easy. SIbotaccess is the simple way a business closes that gap — you declare the actions an agent may take (request a quote, check availability, book an appointment, emergency call-out), publish one small file, and get found and booked by agents, not just by humans.

The open web already has two conventions for the first two layers — robots.txt for crawling and llms.txt for content. SIbotaccess scans both live so you can see where the web stands today, and adds the missing third layer: botaccess.json, which declares the actions an agent may take.

Everything here is built to be read by machines. Every page is server-rendered semantic HTML, every dataset has a machine-readable mirror, and the whole directory is available as JSON and as agent-friendly plain text — the site is its own best example of a property an agent can actually use.