Skip to content

AI Image Generation

Generate custom images and find stock photos right from the AI chat. Built-in, no API keys required.

Proyecta’s AI can generate images during the build process. There are two built-in tools — one for custom AI-generated visuals and one for curated stock photography. No third-party API keys required.

The AI has a generate_image tool that creates custom visuals on demand. Just ask:

  • "Generate a hero image for my landing page — abstract gradient with floating shapes"
  • "Create product mockup images for my e-commerce store"
  • "Make an illustration of a team collaborating for the about page"

Generated images are stored in your project’s file storage and get a permanent CDN URL. The AI downloads them into your public/ directory and references them in your code.

Options: The AI can control size (1024x1024, 1536x1024, 1024x1536) and quality (medium, high).

For curated real-world photography, the AI uses generate_unsplash_image to search Unsplash’s library:

  • "Find a hero photo of a modern office space"
  • "Add stock photos of food for the restaurant menu page"

Unsplash images are hotlinked (not downloaded) and include proper photographer attribution automatically.

For logos and simple illustrations, the AI can generate SVGs directly without calling an image API:

Generate an SVG logo for my app — minimalist, two colors, abstract mark.
Save it to public/logo.svg and use it in the navbar.
  • AI-generated images are uploaded to blob storage and downloaded into your project’s public/ directory
  • Unsplash images are hotlinked from Unsplash’s CDN (no download needed)
  • Runtime-uploaded images (from your app’s users) go through the Files & Media SDK

If you need a specific model or provider for your app’s end users (not the builder), the AI can integrate any image generation service:

  • DALL-E / OpenAI"Add an image generation page that calls OpenAI's images API"
  • Stable Diffusion / Stability AI"Integrate Stability AI with an SDXL model"
  • Replicate"Add Replicate-powered image gen with model selection"

Add your provider’s API key in Environment Variables before asking.