The take home point I had from this article was that the requirements of will determine what you need to build. If this is a startup with minimal resources, the

The take home point I had from this article was that the requirements of will determine what you need to build. If this is a startup with minimal resources, the requirement is that you need to create something fast to see if you can solve a problem. At that point you might take those learnings, throw away the code and build a more extensible and robust version 2. The problem I often see is that people over engineer based on a set of requirements which assumes that the problem you are trying to solve is the right problem to solve.