This is the third of four posts in my series on why not to use executable decision models. Part 1 discussed the difficulty in sustaining business user engagement when using executable models, and part 2 outlined the challenges of reuse and maintenance with executable models.
Reason #3: Analytics and AI
We are using decision models to help companies move to data-driven decision-making. Our clients don’t just want to automate decisions, they want to apply data, analytics, machine learning and AI to those decisions. Most executable models assume that every decision can and should be represented as decision logic. But our experience is that many decisions require a mix of decision logic, human judgment, analytics and AI.
We use decision models as requirements for all these kinds of projects, building a decision model first so we understand the decision-making and only then selecting the appropriate technology(ies). This DecisionsFirst approach focuses on the business problem, not the solution approach and definitely not a specific technology.
We also use the same decision models to orchestrate these different kinds of decision-making, mapping those decision models to the underlying technology (BRMS, Machine Learning platform, predictive analytics workbench, AI platform, etc). This lets SMEs see how these technologies are used together to drive better decisions and lets the business pull all its decision-making technologies together into a single digital decisioning platform.
In the final part of this series I will discuss the concept of a virtual decision hub.