Check in two minutes whether ChatGPT can read your website
One terminal command shows you exactly the HTML that the OpenAI, Anthropic and Perplexity bots receive. If it is empty, your company does not exist for AI.
- Level
- No code
- Done from claude.ai or ChatGPT, without programming.
- Time
- Two minutes
- Published
What it costs
Zero. The command is free and the analysis with the prompt fits in the no-cost plan of claude.ai.
guided mode // I ask, you answer
ready to start
Do it with me, step by step
This recipe needs no details from you: I will walk you through the 5 steps, one at a time. Like a Thermomix: you just answer and move on.
Who it is for
Any company with a website, and especially those with a modern site built in React, Vue or Angular.
The problem
The crawlers that feed ChatGPT, Claude and Perplexity do not run JavaScript: they read the HTML the server delivers and leave. If your site is an application that draws itself in the browser, that HTML may be empty. It looks perfect to people and invisible to the machines that answer questions today.
What changes
You know for certain, with evidence you can show, whether your site is readable by AI. If it is not, you have the exact diagnosis to request the fix: static rendering of the content.
Ingredients
- A terminal (PowerShell on Windows; Terminal on Mac)
- Your site's address
- Nothing else
Step by step
- 1
Open the terminal and request your site as if you were GPTBot
Run: curl -A "GPTBot" https://yourdomain.cl. The -A parameter makes you identify with the same name the OpenAI bot uses, so you see exactly what it sees.
- 2
Look for your content between <body> and </body>
If your headings, text and descriptions appear there, your site is readable. If you only see something like <div id="root"></div> or <div id="app"></div>, the server is delivering an empty container that gets filled later with JavaScript.
- 3
Count the readable text
Copy what is between the body tags, strip the HTML tags and count the characters. On oladigital.cl the result before the fix was 2 characters. After the rebuild, more than 6,000.
- 4
If it is empty, ask for static rendering
The technical solution is called static generation or server-side rendering: the HTML arrives complete. Frameworks like Astro, Next.js or Nuxt do it. You do not need to redo the design, you need to change how it is delivered.
- 5
Also check robots.txt and llms.txt
Open yourdomain.cl/robots.txt and confirm it does not block GPTBot, ClaudeBot or PerplexityBot. Then create yourdomain.cl/llms.txt: a text file that summarizes your site for the models.
The prompt
Replace what is in brackets with your company details, or use the guided mode above: it asks you and fills it in for you.
Paste here the HTML returned by the command curl -A "GPTBot" against your site. Analyze it as if you were the bot of an AI engine that does not run JavaScript, and answer: 1. How many characters of visible text are there between <body> and </body>, not counting tags or scripts? 2. Is there an <h1> heading? What is it? 3. Is there structured data (application/ld+json blocks)? What type do they declare? 4. Are there hreflang or canonical tags? 5. With only this HTML, could you explain in one sentence what this company does and where it is? If you cannot, say exactly what is missing. Finish with a one-line verdict: READABLE BY AI or INVISIBLE TO AI, and the main reason. HTML: [PASTE THE CURL OUTPUT HERE]
What usually goes wrong
- ✕Opening the site in the browser to "check". The browser runs JavaScript; AI crawlers do not. You have to look at the raw HTML.
- ✕Trusting that Google indexes it fine. Googlebot does run JavaScript; GPTBot and ClaudeBot do not. They play by different rules.
- ✕Fixing the llms.txt and not the HTML. The llms.txt helps, but if the site content does not arrive, there is nothing to cite.
When not to use this recipe
If your site is WordPress, Shopify, Wix or similar without heavy customization, it almost certainly already delivers full HTML: run the command anyway, but the verdict will most likely be READABLE and you can skip step 4. The real problem is with sites built as applications (React, Vue, Angular) without static rendering.
Source and real case
Digital marketing and web development
This is OlaDigital's own case: the previous site was a React application that delivered 2 characters of readable text to AI crawlers. The rebuild in static Astro took it to more than 6,000, with structured data and hreflang in the raw HTML. The measurement is reproducible with the command in step 1.
Real case, measured and documented:Caso propio: la reconstrucción de oladigital.cl · OlaDigital
Last verified:Our own case, measured with the command in step 1 on oladigital.cl: 2 characters before, more than 6,000 after the rebuild. Anyone can reproduce the measurement.
Share it
If someone on your team could use this recipe, send it in one click. The suggested text copies itself.
Want to apply it in your business? Message Cristián directly.
Telegram