Real-time winter road intelligence powered by an AI agent that fuses government data, weather forecasts, highway camera analysis, and crowdsourced driver reports from social media.
SnowDogs helps Canadian drivers plan safer winter trips by combining:
- official Ontario 511 road conditions and incidents
- Open-Meteo weather forecasts and current conditions
- highway camera retrieval and AI image interpretation
- crowdsourced reports submitted from the app and social channels
- route-level risk scoring and checkpoint-by-checkpoint advisories
- voice safety briefings for hands-free consumption
SnowDogs uses an agentic backend architecture centered on Gemini 2.5 Flash function calling.
- The AI agent receives route checkpoints, crowd reports, and route context.
- It autonomously decides which tools to call (geocoding, official road condition lookup, weather fetches, credibility assessment).
- It iterates through multi-turn tool calls, reconciles conflicting sources, and outputs a structured route briefing with actionable risk segments.
- React + Vite (web) and Expo / React Native (mobile)
- TypeScript end-to-end
- Mapbox GL JS (web) and react-native-maps (mobile)
- Gemini 2.5 Flash with multi-turn function calling
- Cloudinary AI Vision for dual-AI camera analysis
- ElevenLabs Flash v2.5 voice alerts with browser TTS fallback
- Auth0 social login
- Open-Meteo GEM HRDPS 2.5 km weather model
- Ontario 511 REST API for road conditions, cameras, and events
- Backboard.io persistent corridor memory
client/: React + Vite TypeScript frontend (web dashboard)server/: Express + TypeScript backend — routes, agent orchestration, risk scoringmobile/: Expo / React Native app with full route scanning and camera analysis- Root scripts: concurrent local development (
npm run dev)