Website guidesGuide 057 min read

    Can ChatGPT and other AI tools actually read your website?

    By Harry Norris, Aucta AI · Updated 26 October 2026

    Quick answer

    Often, but not always. Most AI tools that visit websites, including ChatGPT's and Claude's, read a page as it first arrives and do not run JavaScript. If your site builds its words in the visitor's browser, they may see an almost empty page. Your robots.txt file can also block them by accident. Both take two minutes to check.

    On this page

    More people now ask ChatGPT, Claude or Perplexity who to call for a job. Those tools can only recommend a business they can read about, and a surprising number of websites are close to invisible to them. This guide explains why in plain English, gives you two quick tests, and covers what to do if your site fails. It is the technical side of being found by AI; for the bigger picture, read what GEO means for a local business. Every website we build is set up to pass both tests from day one.

    How AI tools read a website

    AI companies send automated visitors, called crawlers, to read web pages. They do three different jobs. Some collect pages to train the AI models. Some build a search index, so the AI can quote and link to your site in its answers. And some open a page live, when someone asks the AI about a particular business or link.

    The crawlers behind ChatGPT, Claude and Perplexity have one thing in common: they read the page exactly as your server sends it. Unlike a person's web browser, they do not wait around to run the code that builds the rest of the page.

    Built on the server, or built in the browser?

    There are two ways a website can deliver a page. Think of it as the difference between a printed leaflet and a flat-pack kit.

    • Built on the server (the leaflet). The page arrives finished, with every word already in it. People, Google and AI crawlers all see the same thing straight away.
    • Built in the browser (the flat-pack kit). The page arrives almost empty, with a set of instructions written in JavaScript. Your browser follows the instructions and assembles the page in a split second, so it looks fine to you. A crawler that does not follow the instructions just sees the empty box.

    When Vercel, the company behind Next.js, studied AI crawlers across its network, none of the major ones ran JavaScript. Google's and Apple's crawlers were the exceptions. So for most AI tools, a flat-pack page is a page with nothing on it.

    Which websites have this problem?

    Most WordPress, Wix and Squarespace sites send their words in the page, so they usually pass. The problem is more common on:

    • custom websites built as JavaScript apps, where the whole site is assembled in the browser;
    • some sites made with AI app builders, which often produce this kind of app by default;
    • parts of otherwise fine sites that load afterwards, such as price lists, menus or reviews pulled in from another service;
    • important details written inside images, such as a banner with your phone number or the areas you cover.

    Test 1: can the words be seen without JavaScript?

    This shows you roughly what an AI crawler sees. It works best on a computer.

    1. Open one of your pages in Chrome, Edge or Firefox.
    2. Right-click an empty part of the page and choose View page source (or press Ctrl+U, or Cmd+Option+U on a Mac).
    3. Press Ctrl+F (Cmd+F on a Mac) and search for a sentence you can see on the page, such as your main heading or one of your services.

    If the sentence is there, your words are in the page and AI crawlers can read them. If it is missing, and the source is mostly code, your page is being built in the browser, and most AI tools will see very little of it.

    Two-minute check

    Is your website blocking AI tools?

    Every website can have a robots.txt file that tells crawlers what they may read. Open yours, copy everything in it and paste it below. Nothing is sent anywhere; the check runs in your browser.

    Open robots.txt

    If the page says "not found", you have no rules and every tool below is allowed.

    Blocking training crawlers, such as GPTBot, ClaudeBot or Google-Extended, is a fair choice if you do not want your words used to train AI models, and it does not stop you appearing in their search answers. Blocking the search crawlers is different: if OAI-SearchBot or PerplexityBot cannot read your site, ChatGPT and Perplexity cannot recommend it. Website builders and security plugins sometimes add these blocks without anyone noticing.

    Readable is not the same as understood

    Passing both tests means AI tools can read your site. Whether they recommend you depends on what they find. The sites that do well tend to:

    • say plainly what the business does, where, and for whom, near the top of each page;
    • have a separate page for each main service, rather than one page listing everything;
    • use the same business name, address and phone number everywhere, including your Google Business Profile;
    • answer the questions customers actually ask, in their own words;
    • show real reviews and real examples of work.

    For the rest of the picture, including reviews and mentions on other sites, see how to get your business recommended by ChatGPT. None of this needs special technology. It is the same clear, useful writing that works for people. The difference is that AI tools are less forgiving of vague pages than people are. Choosing how your site is built matters here too; our guide to Next.js and WordPress for a small business compares the two most common options.

    From our work

    Our own website is built with React, which by default builds pages in the browser: exactly the problem this guide describes. So every time we publish, each page is built in advance into a finished page, and an automatic check flags any page that is missing its main heading or description.

    The websites we build for clients use Next.js, which sends finished pages from the server as standard, so they pass both tests without any extra steps.

    If your site fails

    • Blocked in robots.txt. The easiest fix. Ask whoever looks after your site to remove the lines blocking the search crawlers you want. It takes minutes.
    • Words missing from the page source. Ask your developer whether the site can be "server-rendered" or "pre-rendered", which means building each page in advance. Some sites can add this; others need rebuilding.
    • Key details in images. Put your phone number, services and areas in normal text, and keep the image for decoration.
    What we would do

    A custom website from £97 a month, yours to keep.

    Every website we build sends finished pages from the server and is written to be understood by Google and AI tools. £97 a month with nothing upfront, hosting and small updates included, and yours to keep after 12 months.

    Common questions

    Can ChatGPT read my website?

    Usually, if the words are in the page when it first arrives and your robots.txt file does not block OpenAI's crawlers. OpenAI's crawlers do not run JavaScript, so a site that builds its words in the visitor's browser can look almost empty to them. Both are quick to check using the two tests in this guide.

    Should I block GPTBot?

    It is your choice. GPTBot collects pages to train OpenAI's models; blocking it does not stop your site appearing in ChatGPT search, which OpenAI controls with a separate crawler, OAI-SearchBot. If you want to be recommended by ChatGPT, do not block OAI-SearchBot.

    Can Google read websites built with JavaScript?

    Google's crawler can run JavaScript, so it can usually see those pages, and Google's AI Overviews draw on the same index. A page that arrives complete is still quicker and safer for Google, and it is the only kind most AI chat tools can read.

    Does schema markup help AI tools understand my site?

    Schema markup is structured information in the page's code, such as your business name, address and opening hours. It helps search engines, and it gives AI tools clean facts if they read it. How much each AI tool uses it directly is not published, so treat it as useful rather than a guarantee.

    How do I make my website readable by AI?

    Make sure each page arrives with its words already in it, do not block the search crawlers you want to appear in, and write each page so it clearly says what you do, where, and for whom. Put key facts in text, not images.

    Cookie Preferences

    We use cookies to enhance your experience and analyse how visitors interact with our website. Read our Privacy Policy.