AI visibility

OpenAI

Can ChatGPT read your site?

Three OpenAI user agents decide whether ChatGPT can find, cite and read your site.

Reads robots.txt and llms.txt. Nothing stored.

How it reads the web

OpenAI documents three user agents. GPTBot collects pages for training. OAI-SearchBot builds the index behind ChatGPT search, which is where citations come from. ChatGPT-User fetches a page live when someone asks about it. Each one respects its own robots.txt rule, so a site can be open to one and closed to another. None of them is a browser: what they read is the HTML the server returns.

Documented user agents

  • OAI-SearchBotchecked
  • ChatGPT-Userchecked
  • GPTBotchecked

What your site needs

Four things that decide whether ChatGPT can use your pages

  1. 01

    Allow OAI-SearchBot and ChatGPT-User

    Blocking GPTBot is a training decision. Blocking the other two removes you from answers and citations.

  2. 02

    Real text in the initial HTML

    An empty application shell that fills in with JavaScript reads as an empty page.

  3. 03

    A canonical on your production domain

    If the canonical still points at a builder preview domain, that is the address ChatGPT associates with the content.

  4. 04

    Structured data and a clear entity

    JSON-LD that says what the site is gives the model something to quote instead of guessing.

Why builder sites fail here

  • Client-rendered React shells with no server output
  • A robots.txt that disallows everything on a preview host and was copied to production
  • The builder default title on every route

BulkAudits finds these on a live site, then opens a pull request on a new branch with the fix. You merge.

Audit my site

Questions about ChatGPT

Does blocking GPTBot hide my site from ChatGPT?
Not from answers. GPTBot is the training crawler. ChatGPT search uses OAI-SearchBot for its index and ChatGPT-User for live fetches, and each has its own robots.txt rule.
Does ChatGPT run my JavaScript?
Do not count on it. OpenAI describes its agents as fetching pages, not rendering them. Put the words that matter in the HTML the server sends.
What does BulkAudits check?
Whether all three OpenAI agents are allowed in robots.txt, whether sampled pages carry readable text before scripts run, canonicals, structured data, entity clarity, headings, the sitemap, robots configuration and llms.txt. Then it opens a pull request for what can be fixed in code.

Other assistants

Every check above is part of the nine-point AI readiness score in a full audit.