Propulse

# Propulse: AI-Powered Proposal Generation System

Propulse is a multi-agent system that leverages AI to generate high-quality proposals based on user prompts and RFP documents. The system uses vector databases to retrieve relevant context from past RFPs and proposals, ensuring generated content is both accurate and contextually appropriate.

## 🏗️ Architecture

### System Architecture
```mermaid
graph TD
    subgraph "Frontend Layer"
        UI[Streamlit UI]
        Upload[Document Upload]
        Preview[Proposal Preview]
    end

    subgraph "Backend Layer"
        API[FastAPI Service]
        Auth[Authentication]
        Cache[Redis Cache]
    end

    subgraph "Agent Pipeline"
        R[Retriever Agent]
        W[Writer Agent]
        V[Verifier Agent]
    end

    subgraph "Storage Layer"
        VDB1[Vector DB - RFPs]
        VDB2[Vector DB - Proposals]
        DB[(PostgreSQL)]
        GCS[Cloud Storage]
    end

    UI --> API
    Upload --> API
    API --> Auth
    API --> Cache
    API --> R
    R --> VDB1
    R --> VDB2
    R --> W
    W --> V
    V --> API
    API --> Preview
    API --> DB
    API --> GCS
```

### Workflow Diagram
```mermaid
sequenceDiagram
    actor User
    participant UI as Frontend
    participant API as Backend
    participant R as Retriever
    participant W as Writer
    participant V as Verifier
    participant DB as Databases

    User->>UI: Upload RFP/Enter Prompt
    UI->>API: Submit Request
    API->>R: Get Relevant Context
    R->>DB: Query Vector DBs
    DB-->>R: Return Matches
    R->>W: Context + Prompt
    W->>V: Generated Proposal
    V->>API: Verified Content
    API->>UI: Return Proposal
    UI->>User: Display Result
```

## 📁 Detailed Project Structure

```
Propulse/
├── backend/                 # FastAPI backend service
│   ├── agents/             # Agent implementations
│   │   ├── retriever/      # Retriever agent logic
│   │   │   ├── __init__.py
│   │   │   ├── agent.py
│   │   │   └── utils.py
│   │   ├── w