Robinsight: route calculations from days to minutes
Runs that took days or weeks now finish in minutes, freeing capacity to take on new clients
"[ROBINSIGHT CLIENT QUOTE: 1-3 sentence pull quote from founder about impact]"— [Name], [Role], Robinsight Ltd
The problem
Robinsight describes itself as New Zealand's most experienced telematics analytics organisation. It turns raw vehicle telematics into decision-ready intelligence for government, academia and industry, including NZ Transport Agency Waka Kotahi, the Ministry of Transport and Auckland Transport. By its own figures, its data covers more than 120,000 connected vehicles in New Zealand.
Turning that data into answers means calculating routes for very large numbers of trips. Those calculations ran on desktop machines and took days or weeks to complete.
That runtime capped the business. Projects had to queue behind one another, runs needed watching, and a failure meant starting again. The team could only take on as many projects as their machines could get through.
What I built
A cloud data-processing pipeline that uses low-cost compute only while calculations run, and shuts down in between. Work runs in parallel instead of in a queue, and the system handles failures without anyone stepping in.
Robinsight pays for compute only while a calculation is running, and several projects can run at once.
The outcome
Runs dropped from days to minutes, with far greater reliability.
Processing stopped being the limit on the business. The team now has the capacity to take on new clients, and the founder considered the price trivial next to the capacity it freed up.