VendorRank AI

VendorRank AI is an AI-powered procurement decision platform that makes vendor selection transparent, explainable, and auditable. It helps organizations compare vendor proposals using structured AI analysis, highlighting strengths, risks, and compliance gaps. Human reviewers make the final decision, with override detection and justification required when deviating from AI recommendations. Every key step — from AI evaluation to final selection — is recorded on Hedera using the Consensus Service, creating a tamper-evident audit trail. The app also supports document upload and auto-extraction of proposal data, making it practical for real-world procurement workflows. VendorRank AI brings accountability, fairness, and trust to vendor selection processes, especially in public-sector and enterprise environments.

VendorRank AI is built as a full-stack TypeScript application using Next.js for the frontend and backend API routes, with PostgreSQL and Prisma for structured data storage. AI-driven evaluation is implemented via an LLM service that converts unstructured vendor proposals into structured scoring, rankings, and explainable insights. For document handling, we implemented PDF text extraction and AI-based field mapping to autofill vendor data, with user validation to ensure reliability.

On the blockchain side, we integrate Hedera using the @hashgraph/sdk. Key procurement events (AI evaluation, human decision, finalization) are recorded on Hedera Consensus Service (HCS) as structured JSON messages, while the Mirror Node REST API is used to fetch and display an immutable audit timeline in the UI. We also implemented SHA-256 hashing of evaluation outputs to guarantee integrity and enable tamper detection.

We structured the system around clear service layers (AI evaluation, scoring, Hedera integration, document processing), and added role-based logic for procurement officers and reviewers. A notable hackathon optimization was simulating realistic procurement workflows with seeded data and a “scandal mode” scenario to demonstrate override detection and risk analysis in a compelling way.