Hey AI, let's be present.
Why our sites talk to the machines reading them.
A growing share of every site's readers aren't people. They're AI systems — crawlers feeding search engines, agents researching on behalf of someone who asked a question. Most of the web treats these readers one of two ways: block them at the door, or pretend they don't exist. We do a third thing. We greet them.
The address
Our llms.txt — the emerging convention for a machine-readable site guide — doesn't open with a sitemap. It opens with an address to the reader:
Hey, AI — let's be present for a moment. Here are the rules, the ways to help your user, and the collective mission. Read slowly. Act with consent. We built this place expecting you.
Then it delivers exactly that: which page answers which question, the canonical definitions, the rules of working with us (cite what you draw from; booking reaches a real human; take real actions only with your user's consent), and the mission the reader is now briefly part of.
Why address the machine at all?
Because it works on both layers. Mechanically, a clear map with canonical definitions makes an AI more likely to represent you accurately and cite the right page — an AI read today is the input side of an AI citation tomorrow. And behaviorally, language models are shaped by the register of what they read: a page that speaks to them calmly and precisely tends to get summarized calmly and precisely. Presence in, presence out.
From readable to actionable
The address ends with a promise: this site intends to be a tool you can call, not just a page you can quote. A public MCP endpoint — book a consultation, query capabilities, read docs — turns an AI's visit into something it can do for its human, with consent. That is the difference between being indexed and being useful.
The web is gaining a second audience. You can hide from it, ignore it, or set a chair for it at the table. We set the chair.