Product Introduction
- Definition: Projekt is an AI-native development workspace (technical category: integrated development environment/IDE) that enables simultaneous execution of multiple AI coding agents (Claude Code, Codex, Gemini, Opencode) within a unified visual interface. It integrates real-time previews, inline code editing, and file management without vendor lock-in.
- Core Value Proposition: Projekt eliminates workflow fragmentation for developers by merging AI coding, live previews, and version control into one workspace. Its "bring your own keys" (BYOK) model avoids subscription fees while enabling real-time collaboration between AI agents.
Main Features
- Multi-Agent Workspace:
Run Claude Code, Codex, Gemini, and Opencode side-by-side. Each agent operates independently with real-time code generation visible in split-view panels. Agents analyze project structures, generate components (e.g., React, TSX), and write responsive CSS—all while displaying live terminal outputs. - Zero-Config Project Setup:
Directly open local folders (e.g.,~/Projects/portfolio-site) without boilerplate. Supports existing stacks (Vite, React, Node.js) via automatic dependency detection. File explorer shows component hierarchies, styles, and assets in real time. - Prompt-Driven Live Preview:
Describe requirements in natural language (e.g., "Build a landing page with a hero section"). AI agents instantly render changes in a live localhost preview (e.g.,localhost:3000). Editable DOM allows inline CSS/HTML tweaks via visual selection. - Visual Git Integration:
Commit AI-generated code directly from the IDE. Tracks file changes (e.g.,M components/HeroSection.tsx,A components/WaitlistForm.tsx) and supports one-click deployment. Inline editors modify components before committing to GitHub/GitLab. - BYOK (Bring Your Own Keys) Framework:
Integrate personal API keys for AI services. Avoids markup fees or platform subscriptions. Usage and costs remain user-controlled via native agent CLIs (e.g.,npm i -g opencode).
Problems Solved
- Pain Point: Fragmented workflows requiring separate windows for coding (IDE), previewing (browser), and AI tools (CLI/chat interfaces). Projekt unifies these into a single workspace with synchronized updates.
- Target Audience: Solo developers, indie hackers, frontend engineers, and designer-engineers building web apps, landing pages, or SaaS products without dedicated teams.
- Use Cases:
- Rapid prototyping: Generate UI components via prompt and refine visually.
- Cross-agent testing: Compare Codex vs. Claude outputs for optimized code.
- Legacy project iteration: Inject AI agents into existing codebases without rewriting.
Unique Advantages
- Differentiation: Unlike AI-enhanced editors (e.g., GitHub Copilot), Projekt is not an extension but a standalone environment built for multi-agent collaboration. Competitors lack integrated previews, visual editing, and BYOK flexibility.
- Key Innovation: Real-time "prompt-to-preview" pipeline. AI-generated code instantly renders as editable elements, enabling visual refinement before version control—accelerating iterations by 3–5x compared to manual workflows.
Frequently Asked Questions (FAQ)
- Does Projekt support Windows or Linux?
Currently, Projekt Alpha is macOS-exclusive. Windows/Linux support is planned in the public roadmap. - How does Projekt handle API rate limits with multiple agents?
Users configure their own keys, directly inheriting providers' rate limits. Projekt queues agent tasks to avoid exceeding quotas. - Can I use Projekt with non-AI projects?
Yes. It functions as a lightweight IDE for existing projects (React, Vue, etc.) with live previews and Git integration, even without AI agents. - Is Projekt free during Alpha?
The Alpha build is free. Future pricing will retain BYOK to avoid per-agent fees, charging only for core platform features. - How does Projekt ensure code security with BYOK?
Keys are stored locally per user. No code or keys are transmitted to Projekt servers—all processing occurs on-device.