chapter five

5 Solve from a business perspective first

 

This chapter covers

  • Positioning the AI solution as a new way of working rather than as an algorithm
  • Designing how humans and AI can best complement each other
  • Adopting a hypothesis-driven approach to AI problem solving
  • Selecting the best-fitting type of AI for a given problem
  • Testing analytics maturity, problem-analytics fit, and data quality

It is time for a change of mindset. Previously we focused on empathy and discovery to understand needs from different perspectives and to define the problem. We will now focus on solving that problem efficiently. We begin by presenting the solution as a new way of working—a workflow accompanied by clear decision rights—before discussing models or algorithms. Breakthroughs in AI do not come from models alone, however powerful they may be, but from how AI is applied to improve the way work is done. When redesigning workflows in which humans and AI increasingly coexist, it is essential to understand the core competencies of both in order to design the right balance of responsibilities. We propose criteria to help find that balance.

We then operationalize the main problem-solving method that we introduced earlier and will accompany us throughout the AI Road Test: the hypothesis-driven approach. We revisit its fundamentals and explain how to apply it in our specific context.

5.1 An efficiency quest constrained by reality

5.1.1 Increasing gluten production yield with AI.

5.1.2 From monitoring to optimizing

5.2 Positioning the AI solution as a new way of working

5.3 Defining the scope of AI within the workflow

5.3.1 Competency differences: AI vs humans

5.3.2 AI and human strengths

5.4 Describing the new way of working as a hypothesis

5.4.1 Hypothesis-Driven Problem Solving in the AI Road Test context

5.4.2 A short but effective story to describe the proposed solution

5.4.3 Logical arguments to accompany the proposed solution

5.5 Which AI for my use case?

5.5.1 How to group AI algorithms in a structured way

5.5.2 The trade-off between accuracy and interpretability

5.5.3 A practical way forward on AI selection during the AI Road Test

5.6 Matching AI ambition to enterprise analytics maturity

5.6.1 A practical rule to balance AI sophistication and organizational readiness

5.6.2 Reconciliating vision and short-term feasibility

5.7 Is the data ready?

5.7.1 The Key questions that EDA should answer

5.7.2 Bringing actionable conclusions to the EDA

5.8 Communicating the new way of working effectively

5.9 Practical tools for designing a new AI-powered workflow

5.9.1 A simple and practical guide on where to use AI

5.9.2 An AI prompt to support the benchmarking of AI approaches

5.10 Defining the new way of working at Ecoverto