Quick solutions to fix problems and get your AI-powered website redesign back on track
If your preview window stays blank or won't refresh, the most common cause is that the AI is still generating files. The preview will automatically update once all files are created and the development server starts running. You can monitor the progress in the Logs panel at the bottom of the screen.
Sometimes browser security settings can prevent the preview from displaying. Make sure your browser allows iframes and check the browser console for any JavaScript errors that might indicate what's blocking the preview. If you still don't see anything after the generation completes, try manually refreshing the preview using the refresh button in the preview toolbar.
When your website doesn't adapt properly to mobile screens, the issue usually lies in fixed widths or missing responsive breakpoints. The easiest solution is to simply ask the AI to "Make this fully responsive for mobile" and it will update the CSS to use flexible layouts and proper media queries.
For more control, you can request specific changes like testing at particular breakpoints (768px for tablets, 375px for phones) or asking for a mobile-first design approach where the layout is built for mobile first and then enhanced for larger screens. Look for any fixed pixel widths in your layout and ask the AI to convert them to percentages or flexible units.
When the AI doesn't produce what you're asking for, the problem is usually with how the request is phrased rather than with the AI itself. The key to success is being specific and clear about what you want to change. Instead of saying "make it better," try something like "increase the font size to 18px and add 40px of padding around the hero section." The more specific you are, the better results you'll get.
Breaking down complex changes into smaller, sequential steps also helps tremendously. Rather than asking for a complete redesign all at once, tackle one section at a time. For example, first ask to update the header navigation, then move on to the hero section, then the features area, and so on. This approach gives you more control and makes it easier to get exactly what you want.
Providing examples is another powerful technique. If you've seen a design you like on another website, mention it specifically: "Make the layout similar to the pricing section on stripe.com." You can also use design terminology that the AI understands well, like "hero section," "grid layout," "card components," or "responsive breakpoints." This helps the AI understand exactly what design patterns you're looking for.
If the AI still doesn't understand after your first attempt, don't get frustrated—simply try rephrasing your request using different words or approaching the problem from a different angle. Sometimes just rewording your request can make all the difference in getting the results you want.
View example prompts →If the generated design isn't what you had in mind, it's usually because the AI didn't have enough information about your specific preferences. The AI makes reasonable design choices based on modern web design principles, but it can't read your mind about brand colors, typography choices, or specific layout styles you prefer. The solution is to provide detailed design requirements upfront or iterate with specific feedback.
Start by specifying your exact color palette (using hex codes like #3b82f6), font preferences (like "use Inter for body text and Poppins for headings"), and layout style (such as "minimalist with lots of white space" or "bold and colorful"). If you have reference websites you like, share those too: "I want a layout similar to the homepage of linear.app." The more detailed you are, the closer the result will be to your vision.
Remember to make changes one at a time for the best results. Instead of trying to change everything at once, focus on one aspect—maybe start with the color scheme, then move to typography, then layout. This iterative approach helps you maintain control and get exactly the design you want.
When important elements don't appear in the redesign, there are a few possible reasons. First, the AI might not have detected those elements in the original website if you're redesigning an existing site. Some dynamic content or JavaScript-heavy sections can be difficult to capture. Second, if you're building from scratch, the AI will only include what you specifically asked for or what it deemed essential for a basic structure.
The fix is straightforward: explicitly request the missing features. Be very specific about what you need. Instead of saying "add the missing sections," say "add a testimonials section with three customer quotes" or "include a newsletter signup form in the footer." The more specific you are about what's missing and where it should go, the better the AI can add it. If you're redesigning an existing site, you might also want to double-check that the content was actually present on the original page you provided.
If the AI can't analyze the URL you entered, the most common issue is that the website isn't publicly accessible or has restrictions that prevent automated tools from reading it. First, make sure you're using the complete URL including the protocol—type https://example.com rather than just example.com. Some websites work with the www prefix while others don't, so if one version fails, try the other.
Many modern websites use anti-scraping protections that block automated access. If you're trying to analyze a site with heavy security measures, it might not work. In this case, try analyzing a specific internal page instead of the homepage, as sometimes those have fewer restrictions. If the site is behind a login or paywall, unfortunately the AI won't be able to access it—it can only analyze publicly accessible pages.
When you see error messages or broken functionality in your preview, start by checking the Logs panel at the bottom of the screen. This panel shows you detailed error messages from the development server, which can tell you exactly what's wrong—whether it's a missing dependency, a syntax error, or something else. These error messages are invaluable for troubleshooting.
The easiest fix is often to simply tell the AI what you're seeing. Copy the specific error message from the logs and paste it in your chat, then ask "Can you fix this error?" The AI can usually identify the problem and correct it immediately. If you prefer a more general approach, you can just ask the AI to "fix any errors in the code" and it will review the entire codebase for issues. In some cases, you might need to ask the AI to regenerate a specific problematic section rather than trying to patch it.
If the generation process is taking an unusually long time or appears to have stopped, the issue might be with the complexity of your request or your internet connection. Large, complex websites with many pages and features naturally take longer to generate. If you asked for something very comprehensive, consider starting with a simpler version first—get the basic structure working, then ask for additional features incrementally.
Breaking large sites into smaller sections is a much more reliable approach than trying to generate everything at once. For example, start with just the homepage, make sure that works well, then add additional pages one at a time. This not only speeds up individual generations but also makes it easier to troubleshoot if something goes wrong. If the process seems stuck, check your internet connection—a stable, fast connection is important for streaming the AI's responses. When in doubt, refresh the page and try your request again.
If you've tried the specific solutions above and are still experiencing issues, the problem might be with your browser environment rather than the AI tool itself. Browser cache and cookies can sometimes cause unexpected behavior, so clearing them is a good first step. You can usually find this option in your browser's settings under Privacy or History.
Try using a different browser entirely to see if the problem persists—if it works in Chrome but not Firefox (or vice versa), that tells you it's browser-specific. Browser extensions can also interfere with the preview iframe or API calls, so temporarily disable them to rule that out. Finally, network restrictions like corporate firewalls or strict content filters can block some of the necessary requests. If you're on a restricted network, try accessing the tool from a different network or your mobile connection to see if that resolves the issue.
Beyond specific troubleshooting, there are a few general principles that will help you avoid issues and get better results from the start. First, always start simple—begin with basic changes and gradually build up complexity. Trying to do too much at once often leads to confusion and makes it harder to identify what's working and what's not.
Second, test your changes frequently. Don't wait until you've made ten different changes to check the preview. Instead, make one or two changes, preview them, and make sure they look good before moving on. This way, if something goes wrong, you know exactly which change caused the problem.
Finally, be patient with the AI generation process. It can take a minute or two for complex websites to be fully generated, and interrupting the process by refreshing the page or navigating away can cause issues. Let the AI finish its work, and you'll get much better results. Remember that you're creating an entire website—it's worth waiting a bit for quality output.
Sometimes starting fresh with a new session can resolve persistent issues. You can also check our other resources for more guidance.