Signalo — An AI-Powered Civic App Built Solo in 2 Weeks

What problem does Signalo solve?

There are abandoned cars everywhere in Varna — vehicles with expired technical inspections sitting on public streets for months, sometimes years. Most people don't even know they can report them, or what the rules are.

Those who do know face a confusing process: figure out if the inspection is actually expired, write a formal report, find the right email address, attach photos, include location details. It's enough friction that most people just walk past.

I built Signalo to make civic reporting as easy as taking a photo. The app handles everything else — identifying the vehicle, verifying its inspection status against the government database, generating a properly formatted report, and sending it directly to the responsible municipal department.

Why don't citizens report abandoned vehicles?

Reporting an abandoned vehicle in Bulgaria requires knowing several things most citizens don't:

• That vehicles with inspections expired over 3 months are classified as end-of-life vehicles under Bulgarian law (Наредба № 1)
• How to check if a vehicle's inspection is actually expired — through a government website that requires entering the plate number and solving a captcha
• Who to contact — the Repatriation Department (Отдел Репатриране) at the Varna municipality
• What information to include — plate number, vehicle details, GPS coordinates, photos, legal basis

Even for someone who knows all this, the process takes 15–20 minutes per vehicle. For a street with multiple abandoned cars, it's impractical.

The result: abandoned vehicles accumulate, neighborhoods deteriorate, and a solvable problem persists because the reporting friction is too high.

How does Signalo automate civic reporting?

Signalo reduces the entire reporting process to a few taps:

• Take a photo — the app uses AI vision (Gemini Flash, with Groq Llama-4 as fallback) to automatically read the license plate, and identify the vehicle's brand, model, and color
• Automatic verification — the backend uses browser automation (Playwright) to query the Bulgarian Road Transport Agency website, solve the captcha via Google Vision API, and check if the vehicle's technical inspection has expired
• Smart filtering — only vehicles expired over 3 months are included in the report (per the legal threshold). Vehicles expired under 3 months are added to a watchlist — the app sends a push notification when they cross the 3-month mark
• One-tap reporting — the app generates a properly formatted email with all required information (plates, vehicle details, expiry dates, GPS coordinates, Google Maps link, legal citation) and opens it in the user's email client, ready to send to the municipality

The user's only job is to take a photo and confirm. Everything else — plate reading, inspection verification, report formatting, location tagging — is automated.

How did the reporting process change?

What technology powers Signalo?

I built the entire app solo in 2 weeks using AI-assisted coding. I had never touched React Native before — I used AI tools to learn the framework, design the architecture, and write the code as I went.

Key architectural decisions:

• Multi-provider AI cascade — Gemini Flash for plate recognition, with automatic fallback to Groq if rate-limited. Google Vision API as a third fallback for basic OCR. The system stays available even when individual providers have issues
• Browser automation for verification — the government website has no API, so the backend automates a real browser to query inspection status, solve captchas, and parse results. Up to 8 concurrent Playwright instances handle parallel requests
• Smart caching — Redis caches valid inspection results until their expiry date, but never caches expired plates (since a vehicle could get re-inspected at any time)
• No personal data stored — the backend is stateless. Reports go directly from the user's email to the municipality. No user accounts, no signal database, no tracking. Privacy by design
• Offline resilience — failed requests are queued locally and retried when connectivity returns

Where does Signalo stand today?

Signalo is currently in closed testing on Google Play. The app is fully functional — plate recognition, inspection verification, report generation, and watchlist notifications all work end-to-end.

Reports are sent to Отдел Репатриране Варна (Varna Repatriation Department), the municipal body responsible for handling abandoned vehicles.

The goal is to launch publicly and expand to other Bulgarian cities. The underlying system works for any municipality — only the recipient email address changes.

What are the key takeaways from building Signalo?

This project proved something I already believed: you don't need to be a mobile developer to ship a mobile app. I had never touched React Native before. Two weeks later, I had a working product with AI vision, browser automation, push notifications, and offline support — built entirely with AI-assisted coding.

The hardest part wasn't the technology. It was understanding the legal framework, the government verification process, and the municipal reporting structure well enough to automate it.

The AI wrote the code. I designed the system.