Article by Nadine Soyez

 

I often experience this:

When it comes to identifying concrete AI opportunities that would actually make a difference, the organisation is stuck. People wait for someone else to suggest ideas. Teams struggle to see beyond their current way of working. Use case lists from consultants feel generic and disconnected from the real business. And the few initiatives that do get started often fade out because no one is sure whether they are solving problems that matter. The issue is not a lack of AI capability. The issue is that most organisations have no structured way to find where AI would create value. They rely on random inspiration, vendor demos, or copying what competitors claim to be doing. This produces scattered experiments instead of focused impact.

Identifying AI opportunities is a skill. It requires knowing where to look, what questions to ask, and how to separate genuine performance problems from distractions. This newsletter breaks down exactly how to do it: a practical approach to finding AI use cases that connect to outcomes the business actually cares about.

 

 

Start with performance, not with AI

 

The biggest mistake organisations make is asking “where can we use AI?” instead of asking “where is performance constrained, and could AI help?” This distinction matters. When you start with AI, you end up fitting problems to solutions. You look at a tool’s capabilities and search for places to apply them. This creates scattered pilots that may be technically impressive but deliver no measurable business value. When you start with performance, you work backwards from outcomes. You identify where the organisation is slower, more expensive, or less consistent than it needs to be. Then you ask whether AI could address the root cause.

 

Before looking for AI opportunities, define what performance means in your context:

• Speed: How fast can you move from input to decision, from request to delivery, from question to answer?
• Cost: How much does it cost to produce an output, serve a customer, or complete a process?
• Quality: How consistent, accurate, and reliable are your outputs? How much rework happens?
• Capacity: How much can your teams handle without adding headcount? Where are the ceilings?

 

Every AI opportunity should connect to at least one of these dimensions. If you cannot explain how an AI initiative will improve speed, cost, quality, or capacity, it is not a performance opportunity. It is experimentation.

 

 

Find where time disappears into low-value work

 

The highest-impact AI opportunities usually hide in plain sight. They sit in the hours your people spend on work that feels necessary but does not require their expertise. Think about how your teams actually spend their days. Not the strategic work they were hired to do, but the preparation, coordination, and processing that surrounds it. Gathering data from multiple sources. Formatting documents. Summarising meetings. Drafting first versions that will be rewritten anyway. Searching for information that should be easy to find. This is where AI creates immediate and visible value. Not by replacing expertise, but by compressing the work around it.

 

To find these opportunities, ask three questions:

• Where do people spend time on tasks they are overqualified for? If senior analysts spend hours formatting reports, if managers spend afternoons writing routine updates, if specialists spend mornings searching for data they need, these are signals. The gap between what people are paid to do and what they actually do is where AI creates the fastest impact.
• Where does the same work get done repeatedly with slight variations? Proposals that follow similar structures. Reports that pull from the same data sources. Communications that address the same questions. Repetition at scale is an AI opportunity. Standardising and automating these patterns frees capacity for work that genuinely requires human judgment.
• Where do delays create real costs? A proposal that takes five days to prepare delays the deal. A report that takes a week to compile delays the decision. An analysis that requires three people to coordinate delays the project. Map where time loss translates directly into business cost, and you have found a high-value opportunity.

 

 

Look for inconsistency that creates risk or rework

 

Performance is not just about speed. It is also about consistency. When outputs vary depending on who did the work, when quality is unpredictable, when the same question gets different answers from different teams, you have a consistency problem that AI can solve. Inconsistency is expensive. It creates rework when outputs need to be corrected. It creates risk when quality falls below standards. It creates confusion when stakeholders receive conflicting information. And it creates dependency on specific individuals who “know how to do it right.”

 

Signs of consistency problems that AI can address:

• Quality varies by person. Some team members produce excellent work, others produce work that requires heavy editing. The knowledge and standards exist, but they are not evenly distributed.
• Outputs look different every time. Reports have different structures. Proposals have different formats. Communications have different tones. There is no shared standard, so everything is created from scratch.
• Rework is normal. First drafts are rarely final. Reviews consistently catch the same types of errors. People expect to revise multiple times before something is ready.
• Knowledge lives in people’s heads. When specific employees are unavailable, work slows down or quality drops. Expertise is not documented or accessible.

 

AI can encode standards, apply them consistently, and reduce the gap between your best work and your average work. This is not about replacing judgment. It is about ensuring that baseline quality is reliable, so human effort can focus on what genuinely requires thought.

 

 

Prioritise opportunities where you can prove impact

 

Not all AI opportunities are equal, even if they address real performance problems. Some are easy to measure and demonstrate. Others are important but difficult to quantify. Start with the ones where you can prove impact clearly. Provable impact matters for two reasons. First, it builds credibility with leadership. When you can show concrete results, you earn the trust and resources to scale AI further. Second, it builds confidence with teams. When people see that AI actually makes their work better, adoption becomes natural rather than forced.

High-impact opportunities share these characteristics:

• Clear before-and-after comparison. You can measure how long a task takes now and compare it to how long it took before. You can count errors, track rework, or document cycle times. The improvement is visible in numbers, not just feelings.
• Visible to stakeholders who matter. The improvement affects something leadership cares about. A faster sales cycle. A shorter reporting timeline. A reduction in customer complaints. Impact that stays invisible to decision-makers will not earn continued investment.
• Repeatable at scale. The opportunity is not a one-time fix but a recurring process. If AI saves four hours on something that happens once a year, the impact is limited. If it saves thirty minutes on something that happens fifty times a week, the cumulative value is significant.

 

A simple prioritisation approach:

Map potential AI opportunities on two dimensions: business value and measurability. Business value asks how much improvement would matter if you achieved it. Measurability asks how clearly you can demonstrate that improvement. Start with opportunities that score high on both. These are your proof points. They build the case for AI and create momentum for tackling harder problems later. Avoid opportunities that are high value but low measurability until you have established credibility. And discard opportunities that are low value regardless of how easy they are to measure. Easy wins that do not matter are distractions, not progress.

 

 

Validate before you invest

 

Once you have identified a potential AI opportunity, validate it before committing significant resources. Many opportunities look promising in theory but fall apart when you examine the details.

Validation questions to ask:

• Is the pain real and current? Talk to the people who do the work. Are they frustrated by the current process? Do they see the time loss and inconsistency you identified? If the people closest to the work do not feel the problem, AI adoption will struggle.
• Is the data available? AI needs inputs. If the opportunity requires data that is scattered, incomplete, or locked in systems that do not integrate, implementation will be harder than expected. Check data availability before assuming a solution is feasible.
• Will the organisation accept the change? Some opportunities require changes to how people work, who reviews outputs, or how decisions are made. If the organisation is not ready for these changes, even a technically successful AI implementation will fail to deliver impact.
• Can you run a small test? Before scaling, can you pilot the AI solution with a single team, a single workflow, or a single use case? Small tests reveal problems early and reduce the cost of learning.

 

 

What companies should do now

 

If your organisation is struggling to find AI opportunities that actually matter, start here:

• Define what performance means for your business. Be specific about speed, cost, quality, and capacity. Make these the filter for every AI opportunity.
• Map where time disappears into low-value work. Interview teams. Shadow workflows. Find the gap between what people are hired to do and what they actually spend time on.
• Identify where inconsistency creates cost. Look for rework, quality variation, and dependency on specific individuals.
• Prioritise opportunities you can measure and prove. Start where the before-and-after comparison is clear and visible to leadership.
• Validate before you invest. Confirm the pain is real, the data is available, and the organisation is ready.

 

The organisations that get value from AI are not the ones with the most pilots or the most tools. They are the ones that connect AI to performance problems that matter and prove the impact in terms the business understands. AI opportunities are everywhere. AI opportunities that actually change performance are rare. The difference is discipline: starting with outcomes instead of tools, finding where time and consistency are lost, and prioritising what you can measure and prove. Stop asking where you can use AI. Start asking where performance is constrained. That is how AI moves from activity to impact.