SignalRadar

SignalRadar monitors vendor watchlists, finds fresh public risk signals, shows what changed, and drafts the next action before vendor issues become customer issues.

Comment

Inspiration

Vendor risk changes faster than quarterly reviews. A supplier can look fine in yesterday's spreadsheet and become urgent overnight because of an outage, status-page update, advisory, or customer escalation. Procurement, security, and operations teams need a way to see which vendor changed, why it matters, and who should act first.

SignalRadar is built for that gap.

What it does

SignalRadar monitors a vendor watchlist, searches the live web for fresh public risk signals, scores each vendor, and turns the result into a ranked action queue. The dashboard shows portfolio risk, new signals, per-vendor deltas, source-backed evidence cards, and a recommended next action.

The demo's memorable moment is the overnight scenario: Northstar Hosting moves from routine monitoring to critical review after new outage and security-advisory signals appear. Instead of a generic AI summary, SignalRadar shows the evidence, the score movement, and the operational action in one place.

How we built it

The app is a deployed Next.js and React dashboard on Vercel. Nimble powers live-web search and extraction for current vendor evidence. Groq turns the evidence into compact risk summaries and action memos. Tower is represented as the scheduled data-to-AI orchestration layer: the repo includes a Tower-ready Python worker under tower-app/, a sample scan artifact, and dashboard proof that reads the latest pipeline output.

The Domain Roulette hook is signal.repair: the product repairs noisy vendor alerts into source-backed decisions: what changed, why it matters, and who should act next.

Challenges we ran into

The hardest part was making the demo feel real without becoming noisy. Vendor risk can produce many weak signals, so SignalRadar keeps the workflow narrow: rank the watchlist, preserve source links, explain the delta, and recommend one action. Another challenge was making the sponsor integrations judge-legible, so the UI and submission story explicitly surface Nimble live web, Tower data-to-AI orchestration, Groq analysis, and the signal.repair concept.

Accomplishments that we're proud of

SignalRadar is not just a mockup. It is hosted, has live API health checks, supports seeded fallback data for reliable demos, includes scan-to-scan risk deltas, exports action memos, and has a Tower-ready scheduled worker. The project also has a clean 60-second demo video and public screenshots for judging.

What we learned

The strongest AI workflow is not always the flashiest one. For this problem, the winning shape is a focused product that combines fresh evidence, a clear risk movement, and a practical next action.

What's next

SignalRadar could grow into recurring vendor monitoring with Slack or email alerts, procurement/GRC integrations, owner assignment, audit history, and richer Tower schedules that refresh vendor risk datasets automatically.

Built With

Share this project:

Updates