Inspiration

Enterprise vendor risk management costs $50K+/year through tools like Prevalent or ProcessUnity. SMBs and mid-market companies have no affordable alternative. When SVB collapsed, when SolarWinds was breached, when LastPass was compromised — their downstream vendors had no early warning. VendorGuard changes that.

What it does

VendorGuard is an AI-powered vendor risk intelligence platform. Companies add their vendors, and VendorGuard automatically generates AI risk assessments using Groq's LLaMA model, fires real-time alerts based on risk scores, and provides a live portfolio dashboard showing critical, high, medium, and low risk vendors across the entire supply chain.

How I built it

  • Next.js 16 frontend scaffolded with Vercel v0, deployed on Vercel Edge Network
  • AWS DynamoDB (us-east-1) with three tables: vendors_table, signals_table, alerts_table
  • Groq AI (llama-3.3-70b-versatile) for real-time vendor risk summarization
  • Next.js API routes connecting frontend to DynamoDB and Groq
  • Vercel environment variables for secure credential management

Challenges I ran into

  • Migrating from deprecated Groq models mid-build (llama3-8b → llama-3.1-8b → llama-3.3-70b)
  • pnpm lockfile conflicts when manually editing package.json on mobile without a local terminal
  • Structuring DynamoDB schema correctly for efficient querying across org and vendor dimensions

Accomplishments

  • Full real-time data loop: Add vendor → DynamoDB write → Groq AI summary → signals_table → alerts_table → UI display
  • Built entirely on mobile using GitHub web editor and v0.app
  • Production-grade AWS infrastructure from day one

What I learned

DynamoDB's partition key design matters enormously. The deliberate choice of orgId+vendorId composite keys enables multi-tenant architecture from the start.

What's next

  • NextAuth.js multi-tenant authentication
  • Real-time data feeds (Crunchbase, SecurityScorecard API integration)
  • Slack/email alert webhooks
  • Pricing: $49/month starter, $199/month professional

Built With

  • aws-dynamodb
  • groq-ai
  • llama-3.3-70b-versatile
  • next.js-16
  • react
  • shadcn/ui
  • tailwind-css
  • typescript
  • vercel
  • vercel-v0
Share this project:

Updates