Inspiration

We were alarmed by the fact that 31.9% of Indonesia’s roads are damaged, yet citizens face a bureaucratic maze confused by overlapping jurisdictions and ignored by slow, manual paperwork. We built RoadBlock to cut through the red tape and fast-track road repairs using automation.

## What it does

RoadBlock is an AI-powered web platform that simplifies pothole reporting. Citizens upload a photo, and our system automatically verifies the image, determines the road's jurisdiction via geotagging, and routes the report to the correct government agency. It also enforces transparency by requiring contractors to upload an AI-verified "Proof of Repair" before a ticket is closed.

How we built it

  1. Frontend: Built with React and Tailwind CSS for a responsive, interactive user dashboard.
  2. Mapping System: Integrated React Leaflet for real-time geospatial tracking and jurisdiction mapping.
  3. Orchestration: Utilized n8n as our workflow engine to automate data pipelines and connect services instantly.
  4. AI Engine: Powered by Google Gemini to analyze pothole severity, level of urgency, filter fake reports, and validate repair photos.
  5. Database: Structured through Google Sheets, acting as a rapid, lightweight datastore via n8n integration.

## Challenges we ran into

Time Constraints: Delivering a fully working end-to-end prototype within a grueling 30-hour nonstop hackathon timeline.

AI Agent Token Limits: Managing and optimizing context tokens for Google Gemini to ensure efficient image analysis and consistent structured outputs.

System Synchronization: Fine-tuning the orchestration and webhooks between React, n8n, and Google Sheets to handle real-time data sync without losing data packets.

Accomplishments that we're proud of

Successfully building a fully operational, end-to-end event-driven architecture (React > n8n > Gemini > Google Sheets).

Creating a seamless, zero-friction user experience that completely eliminates the need for citizens to understand complex government bureaucracy.

## What we learned

We learned the power of combining Agile low-code orchestration (n8n) with advanced LLMs (Gemini) to build highly functional production-ready AI agents rapidly, proving that infrastructure solutions don't need heavy, rigid backend code to be effective.

## What's next for RoadBlock

  • Live Testing: Launching field tests to validate the platform with real-world users in a live environment.
  • Inter-Agency Collaboration: Partnering directly with relevant government departments and local road authorities to bridge institutional gaps.
  • Continuous Development: Iterating on the platform to continuously improve features until all 3 core bureaucratic pain points are fully resolved:
  1. Fragmented Jurisdiction: Completely eliminating confusion over National, Provincial, or Regency road ownership.

  2. Byzantine Paperwork & Slow Response: Eradicating manual backlogs by replacing them with instant, transparent tracking.

  3. No Real-Time Prioritization: Providing governments with the data visualization needed to instantly dispatch crews to high-risk potholes.

Built With

Share this project:

Updates