🚀 RFP Velocity

AI-Powered RFP → Proposal Automation for FMCG Suppliers

Transform unstructured RFP documents into structured, priced, ready-to-send proposals — in minutes, not days.

---

📘 Table of Contents

1. Overview
2. Key Features
3. System Architecture
4. Tech Stack
5. Setup & Installation
6. How It Works
7. API Endpoints
8. File Formats
9. Future Enhancements
10. Contributors

---

⭐ Overview

RFP Velocity is a minimal yet high-impact solution that automates the entire RFP response workflow for FMCG suppliers.

💡 What It Does

• Reads raw RFP text (from PDF extraction or direct paste)
• Extracts buyer metadata, deadlines, requirements
• Parses line items & quantities accurately
• Matches items to SKUs using hybrid AI (keyword + LLM reasoning)
• Generates pricing using base cost + margin %
• Produces a professional proposal (HTML / PDF / DOCX)

Built to be lightweight, fast, and hackathon-friendly, while delivering real enterprise value.

---

🚀 Key Features

🔍 Intelligent RFP Understanding

• Extracts buyer name, deadlines, summary
• Detects mandatory requirements & disqualification rules
• Handles unstructured/messy text

📦 Line Item Extraction

• Identifies product descriptions
• Extracts quantities, pack sizes, units, and remarks
• Works even when RFP tables are poorly formatted

🎯 SKU Matching (Hybrid Model)

• Token-based lexical matching for high speed
• LLM-powered reasoning for semantic matching
• Confidence scoring for transparency
• Rejects mismatches (ex: beverage ≠ cleaning liquid)

💸 Automated Pricing Engine

• Reads SKU base cost from CSV
• Applies configurable margin %
• Auto-calculates unit & total pricing
• Ensures consistency across proposals

📄 Proposal Generation

• Clean corporate HTML layout
• Industry-standard formatting
• Supports HTML / PDF / DOCX export
• Ready for client submission

---

🏗 System Architecture

┌─────────────────┐
│   Frontend       │  HTML + CSS + JS
│  (PWA-like)      │
└────────┬────────┘
         │
         ▼
┌─────────────────────────┐
│       Backend API        │ Express.js
│ - SKU Upload             │
│ - RFP Analysis           │
│ - Line Item Extraction   │
│ - SKU Matching           │
│ - Pricing Engine         │
│ - Proposal Generation    │
└────────┬────────────────┘
         │
         ▼
┌──────────────────────────┐
│   Groq LLM (Llama 3)     │
│ - JSON-structured outputs│
│ - Strict prompts         │
└──────────────────────────┘

---

🛠 Tech Stack

Frontend

• HTML5
• CSS3
• Vanilla JavaScript
• Clean enterprise UI design
• Zero dependencies

Backend

• Node.js
• Express.js
• Multer (CSV upload)
• csv-parse (SKU ingestion)

AI Layer

• Groq API
• OpenAI-compatible endpoints
• Llama-3 models for:
  • Metadata extraction
  • Line item extraction
  • SKU semantic matching
  • Proposal formatting

Proposal Export

• HTML (default)
• PDF (Puppeteer)
• DOCX (html-to-docx)

---

📦 Setup & Installation

1️⃣ Clone the Repository

git clone https://github.com/yourusername/rfp-velocity.git
cd rfp-velocity

2️⃣ Install Dependencies

npm install

3️⃣ Create .env File

GROQ_API_KEY=your_key_here
PORT=3000

4️⃣ Start the Server

npm start

5️⃣ Open in Browser

http://localhost:3000

---

🔄 How It Works

Step 1 — Upload SKU Catalog (CSV)

System loads:

• skuCode
• name
• description
• packSize
• category
• baseCost

Stored in-memory for matching.

Step 2 — Paste RFP Raw Text

AI extracts:

• Buyer
• Deadline
• Summary
• Requirements
• Disqualification conditions
• Line items

Step 3 — Run AI Matching

• Hybrid candidate selection
• LLM-powered decision
• Confidence scoring
• Assigns matchedSkuId per line item

Step 4 — Pricing

• Apply margin %
• Auto-calc unit + total price

Step 5 — Proposal Generation

• Clean HTML output
• Professionally formatted
• Export options:
  • .html
  • .pdf
  • .docx

---

🧬 API Endpoints

POST /api/skus/upload-csv

Upload SKU catalog.

POST /api/rfp/analyze

Extract metadata + line items.

POST /api/rfp/:id/match

Runs AI SKU matching with Groq.

POST /api/rfp/:id/generate

Generates proposal + pricing.

POST /api/rfp/:id/download/pdf

Download proposal as PDF.

POST /api/rfp/:id/download/docx

Download proposal as DOCX.

---

📁 File Formats

SKU CSV Example:

skuCode,name,description,packSize,category,baseCost
LEMON_500,Lemon Soda 500ml PET,Lemon drink,500ml,Beverages,18

RFP Input

Raw text pasted into textarea.

Proposal Output

• Clean HTML
• PDF (A4)
• DOCX (Word format)

---

🔮 Future Enhancements

Phase 2

• PDF parsing support
• Dynamic proposal themes
• Multi-user accounts
• Role-based approvals

Phase 3

• ERP Integration (SAP/Zoho)
• Vendor portal submission automation
• Analytics dashboard

Phase 4

• Fully autonomous RFP agents
• Competitive pricing prediction
• Win-rate optimization

---

👥 Contributors

• Tarunya Kesharwani — AI Architecture, Backend, UI/UX, Frontend

---

🙏 Acknowledgements

Thanks to Groq, Node.js, and open-source contributors enabling rapid AI experimentation.

---

📣 Final Note

RFP Velocity demonstrates how Agentic AI can turn long, error-prone RFP workflows into a streamlined, automated, business-winning pipeline.

Feel free to fork, improve, or integrate into real production workflows!