Redesigning the infrastructure behind Rwanda's public transport network.

AC Mobility runs ticketing for Rwanda's upcountry bus network, and was licensing the whole operation from an external vendor for roughly $12,000 a month. Feature requests sat in a shared queue with no committed timeline, and reporting gaps left management working from incomplete information about its own revenue.

An agent with no float cannot sell a ticket. On the vendor platform agents found that out by trying. When they needed more float they photographed a bank deposit slip into a WhatsApp group and waited, and the estimate failed hardest on the days they could least afford to stop selling. That was one of four operational workarounds discovery turned up, and each one marked the edge of what a shared-configuration platform could represent for a network of operators who compete with each other.

I was the sole product designer across two generations of the replacement. V1 shipped in four months and gave AC Mobility ownership of its operations. After thirteen months running it, I proposed and drove V2, rebuilding the model underneath so relationships between buses, stops, routes and operators became assignable rather than fixed, and so float, authority and audit history became part of the interface rather than conditions sitting behind it.

Daily upcountry revenue went from roughly $60K on the vendor platform to $90K on V1 and roughly $130K on V2, on a network that grew from 17 operators to 21. Float failures stopped appearing as a support category, and the $12,000 monthly contract is gone.

V2 also introduced a problem V1 did not have. It became more capable faster than some users became comfortable with its model, and where the correct path costs more thought than the workaround, the workaround wins. That changed how I think about operational software: a capability is not finished when the system can support it, it is finished when the right path is easier than the way around it.

Industry

Transportation & Mobility

Role

Product designer, Researcher

Timeline

Two generations, July 2024-July 2026

Scope

Operations Web, Android POS, Commuter App, Inspector App, USSD, Design System

Scale

21 Operators, 995 Agents, 2,032,930 Tickets/month

Team

V1: 1 PM, 6 Engineers, 1 DevOps V2: 4 Engineers, 2 DevOps