How Proyecta AI Works

Proyecta AI reads and writes code, provisions your data, and publishes your app — all through natural conversation.

Proyecta’s AI can:

• Read and write to your codebase to build features, fix bugs, and change UI
• Provision your data — model your content and seed it in the CMS before the UI is built
• Run setup, build, and data scripts in your development environment
• Create and manage git commits and branches
• Generate images — custom AI visuals and Unsplash stock photos
• Browse the web — navigate pages, click elements, extract data, and take screenshots
• Deploy your app to a live URL with one click
• Track larger pieces of work with milestones during iteration
• Use connectors — call pre-authenticated APIs for Slack, Resend, ElevenLabs, and more

How does the AI build my app?

When you create a new project, the AI starts by gathering requirements — it interviews you about what you’re building, who it’s for, and what features you need. This conversation produces a structured project brief that guides the entire build.

Once the AI understands your goals, it:

1. Plans your data model — the collections and content your app needs — and provisions them in the CMS, seeded with realistic sample content, before any UI exists
2. Builds the UI across your project, wiring your pages to that real data using the app template and your requirements
3. Reviews and refines the result, checking that the build holds together and fixing what it finds
4. Saves a numbered version you can view later in Version History
5. Publishes when you’re ready — your app goes live at a public URL

You see all of this happening in real time through the live preview panel. After the first build, every message continues the app; during that ongoing work the AI can track larger pieces as milestones in the chat panel.

What models does Proyecta use?

Proyecta uses state-of-the-art models by Anthropic (Claude). The model is selected automatically based on the task. Additional model providers coming soon.

Can I share images or files with the AI?

Yes! You can:

• Upload images like screenshots, mockups, or designs by pasting or dragging into the chat
• Annotate screenshots — click the annotate button to draw arrows, rectangles, circles, and freehand marks on a screenshot before sending it to the AI
• Share links to reference external resources
• Use voice — click the microphone button in the chat input for speech-to-text dictation

Can the AI browse the web?

Yes. The AI can open a browser session to navigate pages, click elements, fill forms, extract data, and take screenshots. It uses this to test your app, scrape reference content, or verify external integrations.

What is the difference between the AI Chat and the AI Gateway?

• AI Chat is what you use to build your app. It writes code, creates UI, provisions your data, and publishes.
• AI Gateway lets you add AI features to your published app so your end users can interact with AI — chatbots, content generation, summarization, etc.

Keyboard shortcuts and Command Palette

Press Cmd+K (Mac) or Ctrl+K (Windows) to open the Command Palette — a quick launcher for navigating the builder, switching views, running a Security Audit, and more.

Press Cmd+/ or Ctrl+/ to see all available keyboard shortcuts.

Does the AI notify me when it’s done?

Yes. When the AI finishes a generation, you’ll hear a completion sound and see a desktop notification (if your browser allows it). This lets you switch tabs while the AI works and come back when it’s ready.

What should I do if the AI is confused?

1. Start a new thread (click the history icon in the chat and create a new thread)
2. Be more specific about what you want changed
3. Include a screenshot showing the issue

Proyecta automatically starts fresh threads after 8 hours of inactivity to keep conversation context clean.

Read our Improve accuracy guide for more tips.