Agentic AI System for Automated RFP Discovery, Analysis & Response

This repository contains the implementation of a fully autonomous Agentic AI–driven RFP (Request for Proposal) Automation Platform designed to transform the traditional, manual, and time-consuming RFP workflow for industrial manufacturing enterprises.

The system detects new RFPs, summarizes key requirements, performs technical SKU matching, estimates pricing, and generates submission-ready proposals — all with human-in-the-loop oversight.

---

1. Overview

Industrial B2B RFP handling is still highly manual, leading to:

• Delayed identification of new opportunities
• Hours spent manually matching product specifications
• Lower win rates due to late or incorrect submissions
• Inefficient use of sales, technical, and pricing resources

Since 90%+ of RFP wins depend on timely submissions, automated intelligence becomes essential.

This project delivers a multi-agent AI platform that autonomates the entire lifecycle:

1. Discover new RFPs
2. Summarize requirements
3. Match product specifications
4. Estimate pricing
5. Generate complete RFP responses

---

2. Key Features

Autonomous RFP Lifecycle Management

• Automated scanning of predefined government/enterprise portals
• Extraction and summarization of RFP documents
• Intelligent routing to technical and pricing agents

AI-Driven Technical Matching

• NLP-based requirement extraction
• FAISS-enabled similarity search
• Automatic SKU mapping with “Spec Match %” scoring

Automated Pricing Engine

• Pricing lookup using internal rate tables
• Approximate cost calculations via mock APIs

End-to-End Proposal Generation

• Auto-generated structured RFP responses
• Export to downloadable PDF and Excel formats

Interactive Web Dashboard

• Status tracking
• RFP-level analytics
• Multi-role access for sales, technical, and leadership teams

---

3. System Architecture

Agentic Architecture Components

• Main Agent (Orchestrator) Coordinates the end-to-end workflow and integrates outputs from other agents.

• Sales Agent Crawls/scans URLs, extracts RFPs, and generates summaries.

• Technical Agent Performs requirement extraction and SKU/spec matching.

• Pricing Agent Computes material & service costs.

• Report Generator Produces submission-ready RFP documents.

Technology Stack

Backend

• Python
• FastAPI
• Celery (background agent orchestration)

Frontend

• React
• Bootstrap

Storage

• PostgreSQL
• FAISS (vector similarity search)

AI/NLP

• LangChain
• OpenAI Embeddings
• Tesseract OCR for document extraction

Deployment

• Backend: AWS
• Frontend: Netlify

---

4. Benefits & KPIs

Impact

• 80%+ reduction in manual effort
• 30–40% increase in timely RFP identification
• 20%+ improvement in win rates
• Improved visibility for decision-makers
• Scalable to hundreds of simultaneous RFPs

---

5. Repository Structure (recommended)

/backend
    /api
    /agents
    /models
    /services
    /pricing
    /rpf_scanner
    main.py
/frontend
    /src
    /components
    /pages
    App.tsx
/docs
    architecture-diagram.png
    sample-rfp.pdf
/scripts
    data_ingestion.py
    faiss_indexer.py
README.md

---

6. Getting Started

Prerequisites

• Python 3.10+
• Node.js 18+
• PostgreSQL
• AWS credentials (optional for deployment)

Backend Setup

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

Frontend Setup

cd frontend
npm install
npm run dev

---

7. Future Enhancements

• Integration with SAP/ERP systems
• Fine-tuned domain-specific LLMs for spec extraction
• Automated bidding decision engine
• Multi-language RFP parsing

---

8. Contributors

• Harlee
• Ajay
• Akash
• Akil
• Kavin