How to Find the Perfect AI Use Case
Showing different approaches used in practice
Lately I have seen several companies traveling through their business departments and collecting ideas for AI use cases. As a result they got a lot:
Ideas for new applications
Really specific ideas about AI use cases for their company (based on LLM/GenAI)
Rather “classical” Data Science and ML use cases (still GenAI is not everything)
Use cases or problems to be solved with the classical Data & Analytics stack (ETL, Data Warehouse, Business Intelligence)
So first we can learn, business people do not really differentiate whether or not it is AI (whatever it means) and ideas are here. Second, speaking with the business is possibly a good idea (not only because of running after a trend).
Different Approaches - One Goal
Recently, I've come across the question of “How to find good AI use cases?” several times.
Dilyana Bossenz describes in her post on LinkedIn starting points for ideation:
Problem-driven approach - You have a specific problem to solve
Data-driven approach - Look at your data and what the patterns can deliver
Innovation-driven approach - Creating and supporting new business opportunties
Sol Rashidi describe in her book “Your AI Survival Guide” a rather classical ideation process for a one day workshop:
Explain your “why” defined before to the audience
Explain your data maturity and AI strategy to the audience
Explain the task of the audience (pre-selected people from your company) to brainwrite ideas for AI use cases according to the AI strategy
Designate one person of the group for a later (step 6) to use veto for use cases
Write the ideas without filtering just an open dialog about
Start filtering, discuss the use cases, use the veto-person, result in a cleaned list aligning to your AI strategy
SAP’ e-learning “Applying a Human-Centered Approach to Identify and Define Business AI Use Cases” describe a full design thinking process for AI use cases. Comparable to this topic especially the first “Explore” phase and the “Business AI Explore Workshop” shows us how it can work:
Get started with Business AI - Understand what AI can do for your business
Explore Opportunities - Identify challenges and ideate on how AI capabilities could address them
Define Use Case Ideas - Define selected scenarios in detail including business, technical and ethical perspectives
Compare and Priorize - Prioritize use case ideas based on agreed criteria, such as business value and complexity
Fig. 1: Mural Template for an one day Business AI Explore Workshop
How INFOMOTION (my team/company) is doing it? My colleague Hans Henrik Jorgensen presented with a customer here on YouTube (in german, currently not available) a systematic way we can do it. Initially we see different approaches:
Fig. 2: Approaches for creating Data & AI value use cases
This is the starting point of a step-by-step process to find valuable Data & AI use cases for our customers:
Fig. 3: Data & AI value assessment procedure
What are the approaches you ideate for your AI use cases? Which of the approaches introduced here do you think works best?
Interested in more? My colleague from DAC is currently offering a AI Masterclass about the "5 Pillars of a Successful AI Project" worth doing.