AI-Powered Business Intelligence

Chapter 11. Taking the Next Steps: From Prototype to Production

If you have made it this far and have followed along with the examples and case studies, you should have acquired a solid knowledge of the main AI/ML techniques and their application in the context of various BI scenarios. Congratulations! This is truly a great achievement and puts you in a great position to successfully launch your own AI use case.

In this chapter, we will finally discuss some of the key points of bringing a prototype solution (that’s what we did so far) to production. Conversely, we’ll also discuss why moving a prototype to production might not actually be a good idea. To resolve this apparent contradiction, we need to look at two concepts, originally borrowed from product management: product discovery and delivery.

By the end of this chapter, you should have an intuition about the next steps you should take if your goal is to roll out your AI-powered solution across your organization.

Discovery Versus Delivery

The practical use cases we have discussed so far have been prototypes. You’ve done what product managers call a discovery process. Discovery is about validating value, usability, feasibility, or viability of a product (or a use case).

It’s worth noting that most prototypes are unlikely to survive the discovery phase. That’s the nature of the game: you want to learn fast and fail fast. That’s fine. By now, you shouldn’t have too many resources invested and, ideally, you have other ideas in ...