Key Features

• •Run multiple AI coding agents simultaneously including Claude Code, Codex, Gemini, and Opencode side-by-side in the same workspace with real-time visibility into each agent's activity.
• •Live preview panel updates in real-time as AI agents write code, allowing users to see visual results immediately after describing what they want through natural language prompts.
• •Editable DOM enables users to select any element in the preview and edit it inline, tweak styles visually, and make adjustments through direct manipulation without digging through code files.
• •Prompt-first interface where everything starts with a natural language description, combining AI code generation with immediate visual feedback in a unified workflow.
• •Bring your own API keys with no redundant subscriptions or vendor lock-in, giving users full control over their AI model usage and spending.
• •Single window workspace integrates file explorer, live preview, and chat interfaces without requiring switching between terminal, browser, and editor applications.
• •Open any project folder with no setup, configuration, or boilerplate required, working with existing codebases and project structures immediately.
• •Commit and deploy directly from the workspace with Git integration, allowing users to stage changes and push code when their creations are ready.

Use Cases

• •Building landing pages with hero sections, waitlist forms, and product explanation sections by describing requirements and seeing components generated in real-time.
• •Creating dashboard interfaces and admin panels using multiple AI agents working simultaneously on different components with live visual feedback.
• •Developing design systems and component libraries where users can prompt for UI elements and immediately see how they look and function together.
• •Prototyping web applications quickly by opening any project folder and using natural language to describe features that get implemented with instant preview.
• •Refining existing projects by selecting elements visually and making inline edits to styles and content without navigating complex code structures.
• •Generating complete component sets for features like navigation, forms, and interactive elements through conversational prompts with multiple AI models.

Who is this for?

Projekt targets builders who go from idea to shipped product without waiting on anyone—essentially assuming the user is the whole team. These are designers who engineer, developers who build complete products, and creators who work independently from concept to deployment. The workspace is designed for people who currently switch between terminals, code editors, and browser previews and want an integrated solution that streamlines their entire workflow. It serves users who want to leverage multiple AI coding agents simultaneously while maintaining control over their API keys and avoiding vendor lock-in.