The Vehicle-Routing-Problem is a classic NP-Hard problem - as the complexity of your problem increases, the time it takes to solve grows exponentially. Adding one more destination, or one more vehicle can 10x the computing cost to solve!
While modern computing techniques such as machine-learning can't sidestep this immutable fact, they go a long way to increase the likelyhood of finding very good solutions quickly.
Over the Christmas period I have created my API. It's the product of many months (years?) of research, reading and thought. The kernel of Flow VRP is alive. Having this in place so quickly means that I can focus on making a fantastic user-interface and gives me incredible velocity.
Next up, UI!