DocuSystem

AI-First SaaS application for generating structured Product Requirement Documents (PRDs) and Business Requirement Documents (BRDs) from unstructured notes and ideas.

Architecture

DocuSystem follows a multi-agent workflow architecture:

1. Input: User provides unstructured text/audio transcript
2. Intent Classifier: Analyzes input to determine document type (PRD vs BRD) and required workers
3. Orchestrator: Creates JSON Knowledge Base state and spawns parallel tasks
4. Worker Agents (Parallel):
   • Agent A (Product): Writes User Stories, Problem Statements
   • Agent B (Tech): Writes Schema, API endpoints, Non-functional requirements
   • Agent C (Localization/Quality): Reviews text for "Global English" and clarity (Smartcat readiness)
5. Combiner: Merges outputs into a cohesive Knowledge Base
6. Markdown Generator: Converts JSON Knowledge Base to Markdown (view layer only)

Key Principles

• Source of Truth: JSON Knowledge Base is always the source of truth. Markdown is a view layer only.
• Separation of Concerns: Agent logic (server-side) is separated from UI state (client-side).
• Global English: All AI-generated content follows "Global English" standards for Smartcat compatibility.
• Type Safety: Strict TypeScript with no any types.

Tech Stack

• Framework: Next.js 14+ (App Router)
• Language: TypeScript
• Styling: Material UI (MUI) v5 with custom SaaS dashboard theme
• AI Backend: Vercel AI SDK (Core) for streaming and agent orchestration
• State Management: Zustand for managing JSON knowledge base
• External API: Napkin.ai (placeholder for infographic generation)

Getting Started

1. Install dependencies:

npm install

2. Copy environment variables:

cp .env.example .env

3. Add your API keys to .env:

   • OpenAI API key (or other AI provider)
   • Napkin.ai API key (optional)

4. Run the development server:

npm run dev

5. Open http://localhost:3000 in your browser.

Project Structure

src/
├── app/                    # Next.js App Router pages
│   ├── api/               # API routes (server-side agent logic)
│   └── page.tsx           # Main page
├── components/            # React components
│   └── layout/           # Three-panel layout components
├── lib/
│   ├── ai/               # AI agent implementations
│   │   └── agents/      # Intent Classifier, Orchestrator, Workers, Combiner
│   └── markdown/        # Markdown generator (view layer)
├── store/                # Zustand stores
├── theme/                # MUI theme configuration
└── types/                # TypeScript type definitions

Development

• npm run dev - Start development server
• npm run build - Build for production
• npm run start - Start production server
• npm run lint - Run ESLint
• npm run type-check - Run TypeScript type checking

Next Steps

1. Integrate Vercel AI SDK with actual AI models in worker agents
2. Implement streaming responses for real-time updates
3. Add audio transcript support
4. Integrate Napkin.ai for infographic generation
5. Add export functionality (PDF, Word, etc.)
6. Implement document editing capabilities
7. Add collaboration features