WayFree
Urban Citizens & Commuters: Daily residents facing hazardous urban infrastructure (potholes, open manholes, overflowing garbage bins, broken streetlights). Today, reporting a problem requires dialing bureaucratic call centers or filling complex forms where citizens must manually find coordinates and guess the responsible municipal department. As a result, over 80% of issues go unreported.
Preliminary score
- Value for the user15.5 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI15.0 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing8.5 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility6.5 / 15
Data requirements, running costs and a clear next step.
- Originality4.0 / 10
A meaningful difference from common solutions.
What the judges wrote
- Claude Opus 5.5Preliminary44/100
- Strengths
- Your core citizen flow is coherent and appears to be built: photo upload, address and GPS capture, category and department assignment, a tokenised tracking page for the citizen, and a department page for status updates with an audit trail. Your EXIF write-up is a genuine, specific failure: mobile browsers stripping GPS tags. You fixed it in layers with camera-vs-upload capture, client- and server-side EXIF parsing, geolocation prefetch, and a Leaflet pin-drop fallback with Nominatim reverse geocoding. The deck's tracking screenshot shows the end-to-end result on a real Baku address with map and status history.
- To improve
- Your README repeatedly describes the classification as 'mock AI', and the screenshot shows 1% confidence, so it is unclear whether GPT-4o vision actually drives routing. Show real model outputs on real photos and report accuracy across your 8 categories. Turn the '20+ test photos' into a results table that compares AI category and department against the correct answer, lists the misclassifications, and measures time-to-report against today's call-centre or portal process. Add per-report API cost, data sources and a named pilot partner. Also unify the name: WayFree, CityAssist and ClearCity all appear across your materials.
- GPT-6 AstraPreliminary55/100
- Strengths
- Your submission and tracking screens make the municipal reporting workflow concrete, showing a photo, mapped location, department assignment, and status history. Your account of browsers stripping GPS metadata identifies a specific failure and practical fallbacks rather than merely claiming success. Your proposed one-district pilot provides a focused next step with a clear municipal buyer.
- To improve
- Your next demonstration should establish real GPT-4o classification rather than the mock classification described in the README, then show email delivery and a department update reaching the citizen's tracking page. Your test results should include category-level accuracy, location success rates across devices, and an explanation of the screenshot's 1% confidence value. Compare submission time and routing accuracy against manual intake to establish a measurable adoption benefit. For your pilot, specify department-directory ownership, photo/location privacy controls, and inference, hosting, and geocoding costs.
The team
Agharahim Bakhshiyev
data
Raul Hasanlı
data
Kanan Pashazada
developer
Rinat Aghayev
developer
Gadim GadimovLead
business