CrimeSpots: Live Crime Awareness for Cape Town

South Africa has one of the highest crime rates in the world, but the information people need to stay safe is useless in practice. It sits in WhatsApp groups with hundreds of unread messages, or in official statistics that are a year old by the time anyone maps them. So people decide which route to take and which areas to avoid based on rumour and gut feel. CrimeSpots fixes that. It's a free web app that pulls public crime signals from news sources, uses AI to extract what happened, where and when, then grades every Cape Town suburb red, amber or blue on a live map. You glance at it before a commute or a night out. No reading required. Two design choices matter here. First, privacy: we aggregate to suburb level only, never individuals or addresses, and a dedicated AI pass strips personal information from any user-submitted report before it's stored. That keeps us POPIA-compliant and avoids the anxiety-machine problem apps like Citizen have. Second, access: it's a low-data progressive web app that installs straight from the browser and runs fine on a cheap phone with patchy signal, because that's exactly where the safety gap is widest. The stack is Next.js 15, MapLibre GL JS and Supabase with PostGIS for spatial queries. All AI runs through the Fireworks API on DeepSeek V4: incident extraction, severity scoring, deduplication, and the privacy pass on user reports. Hosted on Vercel. Built solo in Cape Town by Malachi Mathins.