Skip to content
Documentation menu

What agents see

What happens when an assistant visits

An AI assistant fetches your website nothing like a person does. Here is what it actually gets.

When a person opens your website, a browser runs your code: it downloads the page, executes the JavaScript, waits for fonts and images, and assembles what you designed.

An assistant acting for a customer does almost none of that. It requests the page, takes whatever comes back in that first response, and moves on. It is not being lazy — it is doing what it must do to answer someone in a few seconds instead of a few minutes.

The gap this opens

Most modern sites send an almost-empty document first and fill it in with JavaScript a moment later. To a person that gap is invisible. To an assistant it is the whole visit.

So a site can be fast, handsome, and perfectly functional in a browser, and still hand an assistant a blank page. Nothing is broken. Nothing errors. The assistant simply has nothing to report back, so it reports something else — usually a competitor whose page answered.

What it is looking for

Three things, in order of how often they decide the outcome:

  1. Text it can read without running anything — your services, your hours, your area, your prices.
  2. Structure it can trust — machine-readable marks saying this is a business, this is the address, these are the hours rather than leaving it to infer them from layout.
  3. Something it can do — a way to check availability or make a booking that does not require a human to click through a form.

Most sites fail at the first. Almost every site fails at the third.

Check your own in thirty seconds

Turn JavaScript off in your browser and reload your homepage. What remains is roughly what an assistant receives. If the page is blank, or the services and hours have vanished, that is the version of your business an assistant is working from.

navigate openesc close