VendorWatch AI

Enterprises rely on hundreds or thousands of third-party vendors, SaaS tools, outsourcing partners, suppliers, and infrastructure providers. However, vendor risk signals often appear first across the public web before they reach internal risk systems. These signals may include cybersecurity incidents, data breaches, lawsuits, regulatory investigations, service outages, layoffs, financial instability, leadership changes, customer backlash, and reputational issues. VendorWatch AI solves this problem by acting as an autonomous vendor-risk analyst. Given a vendor name and region, the agent plans a risk investigation, generates targeted risk-specific search queries, and uses Bright Data MCP to retrieve live public web intelligence. The system then extracts structured evidence, classifies findings into enterprise risk categories, calculates an explainable risk score, and generates an enterprise-ready Markdown and JSON risk report. The project is designed for procurement, compliance, security, finance, and third-party risk teams that need faster and more evidence-backed vendor monitoring. Instead of relying only on periodic questionnaires or manual research, VendorWatch AI helps teams move toward proactive, live web-based risk intelligence. The current MVP includes a Streamlit dashboard, Bright Data Remote MCP integration, live web investigation through the `search_engine` MCP tool, agent execution trace, structured evidence table, risk score breakdown, downloadable Markdown report, downloadable JSON output, and transparent Bright Data tool usage. VendorWatch AI demonstrates how AI agents can use production-grade web data infrastructure to solve real enterprise risk and compliance problems.