How to make ChatGPT read your Lovable site
Your Lovable site looks perfect in the browser. But GPTBot, ClaudeBot and PerplexityBot don't run JavaScript, and what they receive is HTML with an empty div. To them, your site says nothing.
Painted with JavaScript: AI usually gets an empty page.
Why it happens
Lovable builds single-page apps with React and Vite. The server sends minimal HTML, and it's the browser, running the JavaScript, that paints the copy, the prices and the links.
Google does run JavaScript, with a delay, but AI assistants' crawlers generally don't. They keep the HTML as it arrives, and there's nothing in it to quote.
That's why a Lovable site can show up on Google and still be invisible when someone asks ChatGPT about what you sell.
How to fix it
Check it
Analyze your site with Tarsier: it compares what a person sees with what an AI crawler receives and tells you how much of the text is lost.
Prerender the public pages
Ask Lovable to generate each public route's HTML at build time. The design doesn't change: the text simply comes inside the HTML.
Titles, descriptions and structured data
Each page needs its own <title> and meta description, and the home page a JSON-LD block saying who you are and what you sell.
robots.txt and sitemap
A robots.txt that lets AI bots in and a sitemap.xml with every route, both in the public folder.
Measure again
Publish and analyze again. Parity between what a person sees and what AI receives should get close to 100%.
The prompt to paste into Lovable
My site is built as a single-page app (React + Vite) and the HTML it serves is almost empty: AI crawlers (GPTBot, ClaudeBot, PerplexityBot) don't run JavaScript, so they never see the content. I want every public page to be served already rendered in the HTML, without changing the design: 1. Add prerendering to the build (for example vite-plugin-prerender or an SSG step) for every public route, so each route produces its own index.html with the text, headings and links inside. 2. Give every page its own <title> and <meta name="description">, and add a schema.org JSON-LD block to the home page (Organization or LocalBusiness, plus Product or Offer if I sell something). 3. Create public/robots.txt allowing GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot and Google-Extended, with the Sitemap line. 4. Generate public/sitemap.xml with every public route. 5. Check with curl that each route's HTML contains the main text before any script loads. Don't touch the app logic or the design. When you're done, tell me which routes are prerendered.
Copy it as is. If the analysis finds more problems, the full report includes a prompt written for your site.
Frequently asked questions
Does Google see my Lovable site?
Usually yes, because Google runs JavaScript, though it takes longer to index. AI assistants are different: most of their crawlers only read the HTML.
Do I have to leave Lovable?
No. Prerendering is added to the project you already have; you keep editing in Lovable as usual.
How long until it shows?
What a crawler receives changes as soon as you publish. When an assistant starts quoting you depends on when it comes back, which can be days or weeks.