Three ways to find an AI use case
I’ve noticed that the way we approach use case discovery can be organised into three approaches. Each has its place, and knowing all three will change the kind of opportunities you find. Ultimately, if you only know one way to look, you will only find one kind of opportunity.
The use cases that move the numbers are usually found in one of three ways: by naming a problem out loud, by walking a process end-to-end and watching where it bends, or by holding a current AI capability up against the business and asking, where would this fit? While the first approach is often obvious, the second takes patience, and the third is only possible if someone in the room actually understands what the technology can do today.
The insight: where you look shapes what you find
The valuable use case and the obvious use case are rarely the same one. A use case is an opportunity to make a decision better, faster, or more consistently. Importantly, it needs to be worth it if you implement AI.
The obvious candidates for a use case are not always the valuable ones, and the most valuable opportunities are rarely immediately apparent. The instinct to “make a list of AI ideas” tends to surface the same five candidates every business has heard of. To find the use case that actually fits your business, you need more than one lens.
“The use case you would write down on day one is rarely the use case you would build on day thirty.”
The three approaches
Each approach is suited for a different situation and reveals a distinct kind of opportunity. The skill is knowing which one fits what you have in the room today.
Approach 1
Use-case-first
Start with the problem you can already name
Description
Some problems already have a name, a number, and an owner. The team knows what it hates. The CFO knows what costs too much. When the pain is loud and the boundary is clear, start there.
The risk: You only find the use cases someone already knew to ask for, meaning every shortlist the team creates ends up looking exactly the same.
Use this lens when the pain is loud, named, and owned. A leader or risk officer can already articulate the problem, the cost, and who is accountable. The quickest path to a first win is usually to save money or reduce risk.
Example
A credit team needs to manually check a document. Each check takes 20 minutes, and they need to check 8,000 a month. This is a named problem with a clear owner. Build the solution straight away.
Approach 2
Process-first
Walk the work and watch where it bends
Description
Most opportunities are not announced; they are buried in processes nobody draws end-to-end anymore. Map inputs, decisions, and handoffs. AI candidates appear at the seams: rekeying, classifying, routing, and reviews that almost always say yes.
The risk: Slower to start, but the quieter wins live here.
Use this when the work is messy, and the value is in the seams. Ownership is unclear, multiple teams touch the process, or the obvious candidate seems too good to be true. Walk it end-to-end before you commit.
Example
An insurer maps claim intake end-to-end. Auto-decisioning was already on the list. The quieter win, only visible from the map, is a triage classifier that routes complex claims to a human on day one, not day five.
Approach 3
The AI lens
Apply what AI can do to what the business already does
Description
You cannot ask for a tool you do not know exists. The highest-value use cases often only appear when someone in the room understands current AI, such as extracting structure from messy documents, reasoning over a knowledge base, or scoring millions of events a day.
The risk: Skip this lens, and your shortlist will be five years out of date.
Use this when the team is new to current AI, or the shortlist looks tired. If the same five ideas keep surfacing, the room is missing this lens. New capabilities reveal new use cases, usually the make-money ones.
Example
A retail finance team did not know that an LLM could now read a year’s worth of unstructured supplier emails and flag every quiet price change. This use case never reached a problem list because nobody knew how to ask.
Our approach
We use all three approaches in the order your business needs them. Our Demystifying AI workshop sets the stage utilising all our different “lenses” or approaches discussed above. AI Use Case Discovery workshop “walks through the process” and surfaces the shortlist. Our expertise assists in the AI use case “deep dive”, and we take the chosen use case and build a blueprint. Pick the entry point that matches where you are.
Not sure which lens you need?
AI readiness assessment
The AI Readiness Assessment consists of 12 questions and takes under 5 minutes to complete. It provides an honest read on whether to start with a use case or a process, which approach to take, and what to fix first.
