RouteMate is the simplest pitch in my notebook: most ride-sharing in Bangladesh is expensive, and most commutes are repeated. So why not connect people who already drive the same route every day?
The MVP is free. It's not trying to be Uber. It's trying to be the carpool board in your office hallway, but on a phone and at city scale.
The product
- Routes, not rides. You publish that you drive Mirpur → Gulshan every weekday at 9am. The system finds matches who live and work along that line.
- Social trust. Match scores prefer mutual connections, employer overlap, and prior shared rides. No anonymous strangers in your car.
- Cash-only, settle later. No payment rails in the MVP — neighbours splitting petrol works on trust, not Stripe.
- Built for one corridor first. A real product launches narrow. Dhaka has obvious commuter corridors; we pilot one, learn, expand.
Why I picked the stack
Flutter for the app, Express + TypeScript on the backend, Postgres with PostGIS for geospatial queries. PostGIS is the unsung hero of any location-aware product — once you can express "find me overlapping line strings within 800 metres" in SQL, half the routing logic disappears into the database.
The decisions that mattered
- Match by route geometry, not endpoints. Two people whose endpoints differ but whose paths overlap by 70% are the right match — endpoint-based matching misses them.
- Recurring is the default. One-off rides are a nice-to-have. The whole product is built around I do this every Tuesday.
- No surge. The MVP is free. There is nothing to surge.
The cheapest, safest, most underused mobility asset in Dhaka is empty seats in cars that are already going where you want to go. RouteMate is just trying to make that visible.
What I'd watch in the pilot
The metric I care about isn't downloads. It's repeat matches — the same two people sharing a ride three weeks in a row. If I see that, I have a community. If I don't, I have a download counter.