The three questions we ask before any AI engagement

The hardest part of building AI systems isn’t the engineering. The hardest part is deciding what to build. We see too many teams start with a model and look for a problem, resulting in expensive prototypes that never see production.

Before we write a single line of code, we ask three questions: Does this solve a problem that is currently bottlenecking the business? Can we evaluate the system programmatically? And if we build it perfectly, will the users actually change their behavior to adopt it?

If the answer to any of these is no, we don’t build it. Technical tourism is expensive. We prefer to build boring systems that generate actual margin.