QuotAgent

Inspiration

Industrial RFQs are complex, lengthy, and high-stakes. Sales teams must interpret technical specifications, detect hidden contractual risks, calculate pricing under pressure, and respond quickly — all while protecting margin.

We were inspired to build a system that acts like an intelligent proposal analyst — one that reads RFQs, detects risks, optimizes pricing, and assists negotiation before a human even reviews it.

What it does

It:

Extracts structured data from unstructured RFQs Identifies mandatory vs optional requirements Flags hidden risk clauses Assigns a risk score Dynamically adjusts pricing Recommends optimal product bundles Drafts negotiation-ready responses Tracks outcomes to improve future decisions It transforms a slow, error-prone workflow into a fast, intelligent, and scalable process.

How we built it

Frontend: React + TypeScript for RFQ upload and result visualization Backend: Node.js + Express to orchestrate agents AI Layer: Amazon Bedrock for reasoning tasks such as requirement understanding, risk detection, and negotiation drafting Vector Retrieval: ChromaDB for semantic similarity search across past RFQs

Business Logic: Rule-based pricing and risk scoring for deterministic consistency

Challenges we ran into

Designing a clean agent orchestration flow under hackathon time constraints Simulating realistic risk scoring without historical enterprise data

Accomplishments that we're proud of

Built a working multi-agent architecture in limited time

Successfully integrated Amazon Bedrock for intelligent reasoning Implemented risk-adjusted dynamic pricing logic Designed a negotiation simulation loop Created a scalable architecture aligned with enterprise systems

What we learned

LLMs are powerful for reasoning, but deterministic logic is essential for financial accuracy Clear agent separation improves system scalability Retrieval (RAG) significantly improves contextual intelligence Balancing automation with explainability is critical for business adoption

What's next for QuotAgent

Integrate real enterprise RFQ datasets Implement advanced risk-weight learning models Add real CRM integration Enhance vector-based historical deal intelligence Deploy fully serverless on AWS for enterprise scalability

Built With

• amazon
• amazon-web-services
• bedrock
• chromadb
• gemini
• react