Finding ai opportunities that create real business value

Most organisations do not have an AI tool problem. They have a process, focus and adoption problem.

The challenge is deciding where AI can genuinely improve the way work gets done. The strongest opportunities solve a meaningful problem, contribute to a measurable business outcome and can be tested in practice before deciding whether to scale, adapt or stop.

ai ideas are everywhere. the challenge is finding the ones that matter.

It is Monday morning. The management team is discussing AI again. Marketing is experimenting with Copilot. Customer Service has started a Claude pilot. IT is exploring AI agents. Someone from Sales has found another tool that could save the team hours of work.

There is certainly no shortage of activity.

And yet, when the conversation turns to results, things become less clear. Has customer onboarding become faster? Are we handling more work with the same people? Have error rates gone down? Are customers noticing a real difference?

“We don’t have an AI technology problem; we have a process and adoption problem.”

I hear versions of this in conversations with clients. Another one is just as telling:

“Everyone is experimenting with ChatGPT, but our core metrics and daily friction points haven’t changed at all.”

AI may already be saving people time. Someone prepares a meeting faster. Another person summarises a document in minutes. But zoom out from the individual task and the same handovers, approval delays and fragmented systems may still be there.

People are getting faster. But the organisation is not necessarily getting better.

McKinsey’s 2026 global AI survey makes that gap visible: 80% of respondents say AI has improved their individual productivity, while 37% report a positive contribution to enterprise EBIT. Only about 6% meet McKinsey’s definition of AI high performers.

Related Continno perspective: Organisational change becomes real when everyday work changes

Sketchnote showing a team exploring AI ideas and tools while questioning their impact on business value, capacity, errors and customer experience

The Technology-First Trap keeps AI on the surface

Imagine another familiar situation. Someone sees an impressive AI demo. The question quickly becomes: “Could we use this in our organisation?” A tool is selected. A pilot starts. A team is asked to find some use cases. Only later does someone ask what problem they were actually trying to solve.

That is the Technology-First Trap: starting with what AI can do and then searching for somewhere to apply it.

“If you bolt AI onto a broken workflow, you just get a broken workflow that runs faster.”

Take a customer application that takes ten days to process. AI reduces one 30-minute document-reading task to five minutes. Useful. But what if the application then waits a day for approval, missing information creates rework and three departments still touch the case?

You have automated a task. But have you improved the overall outcome?

RAND’s research into AI project failure highlights the pressure on managers to ‘do something’ with AI and the recurring problem of poorly understood business problems.

McKinsey found that workflow redesign had the strongest relationship with EBIT impact among 25 organisational attributes it studied, yet only 21% of respondents using generative AI said their organisations had fundamentally redesigned at least some workflows.

Explore Continno’s approach to creating more business impact with AI, where the focus is on orchestration rather than another standalone tool.

Sketchnote showing AI reducing document reading from 30 to 5 minutes while the end-to-end customer application process remains slow due to waiting, missing information, rework and handovers

the best place to start is not with ai, but with work that needs improvement.

Instead of asking “Where can we use AI?”, ask: “Where do we need to improve the way we work and what role could AI play?” That change in question shifts the conversation from technology to value.

Before looking for AI use cases, get clear on four things:

What business outcome needs to improve? Think lead time, quality, cost, revenue, capacity, risk or customer experience.
Where is the friction today? Look for waiting, rework, repetitive decisions, manual transfers, searching and unnecessary handovers.
Who experiences that friction? A customer, planner, analyst, service agent, manager or another clearly defined group.
What would meaningful improvement look like? Define the baseline and the result you want before designing the solution.

I saw the value of this approach in practice when working with Unilever Foods Europe. Together with internal AI champions we interviewed employees to understand where repetitive, data-driven and coordination-heavy work was costing time and energy. The team then prioritised opportunities, checked feasibility with Business and Data & Tech, and turned the strongest into experiments.

The principle is simple: AI is the technology

Real business impact comes when people, processes, existing technology and AI perform together, like an orchestra.

AI Opportunity Workshop: Finding the right AI use cases

An AI Opportunity Workshop helps an organisation decide which AI opportunities are actually worth pursuing within a defined scope. This could be a business function such as Marketing or Finance, or a specific product or service.

The goal is not another inspiration session ending with 87 sticky notes. It is to leave with a shortlist of well-framed AI opportunities, grounded in real business needs, user problems and the reality of how the work gets done.

The approach combines the AI Problem Framing method developed by John and Dana Vetan of the Design Sprint Academy with workshop practices I have used with clients such as Unilever. Together, these elements form a practical process that can typically be completed in a well-prepared interactive one-day workshop.

The workshop brings together the people needed to look at the opportunities from different perspectives: business and domain experts, people close to the actual work, someone who understands AI capabilities and constraints, and a decision-maker who can connect choices to business priorities.

Preparation starts before the workshop.

Good preparation makes the workshop much more effective. Existing AI ideas, relevant business goals and KPIs, and available customer or process insights should ideally be gathered beforehand. This allows the workshop itself to focus on understanding the challenges, exploring opportunities, making choices and creating alignment rather than collecting information.

That is why we start with a short project kick-off about one week beforehand. We align on the scope, business challenge and desired outcomes, explain the approach and the next steps.

After the kick-off, the team starts the preparation. Participants interview selected customers, employees or key users to uncover needs, pain points and everyday challenges. They also collect existing AI use case ideas from across the organisation.

To keep this preparation practical and efficient, participants are supported with simple templates, interview guides, AI-assisted interview recording and prompts to help capture and structure the insights. The idea is to gather enough useful input to make better decisions on the day.

Sketchnote showing the AI Opportunity Workshop process from preparation and existing ideas to business goals, customer and employee needs, workflow context and prioritised AI opportunities.

what happens during the one-day workshop?

During the workshop, the team looks at AI opportunities through different lenses: business, people and their work, and technology.

1. Start with the ideas already on the table

Most organisations already have AI ideas, requests or experiments in circulation. We start by bringing them together.

They are the starting point, not commitments to build something. The aim is to make them visible and create a shared view of what is already being considered.

2. Understand the business goals

Next, we apply the business lens. Which goals, priorities and KPIs matter most? What is getting in the way of achieving them?

We map the existing AI ideas against these priorities. Ideas with a clear contribution move forward; others may be parked. This creates a shared business rationale for where to focus and prevents disconnected AI experiments.

3. Understand who you are solving for

Now we shift to the people experiencing the problem. This could be a customer, employee or another internal stakeholder.

We identify the group most relevant to the challenge and explore what they are trying to achieve, what matters to them and which problems are most important to solve.

Using insights from the interviews and existing research, we create a simple proto-persona. This helps the team build empathy and focus on the pain points that really matter.

AI Opportunity Prioritisation – From AI opportunity prioritisation sketchnote showing how ideas are evaluated for business viability, customer desirability and feasibility to create a shortlist of high-potential AI use cases.

4. Understand the work and explore where AI could make a difference

Next, we look at the context in which those problems occur.

We map the relevant customer journey, workflow or service process to understand how the work happens today. Where are the delays, repetitive tasks, difficult decisions or other friction points?

We turn the most important friction points into opportunities and explore where AI could make a meaningful difference.

Existing AI ideas are mapped onto the actual workflow and challenged against what we have learned. Some become stronger, others may disappear, and new opportunities emerge.

The difference is important: AI is now connected to real work and real problems rather than starting with the technology.

5. Choose what is worth pursuing

Finally, we bring together the strongest existing ideas and the new opportunities discovered during the workshop.

We evaluate them from three perspectives: business value, pragmatic feasibility and data availability and customer desirability. This moves the discussion beyond opinions towards a more evidence-based choice about which opportunities deserve further investment.

The result is typically a shortlist of three to five opportunities. The strongest are captured in an AI Use Case Card connecting the target user, problem and friction point, desired outcome, AI intervention, expected value and the business goal it supports.

how does the AI workshop team look like?

Finding valuable AI use cases requires more than business knowledge or AI expertise alone. AI opportunities sit at the intersection of desirability, viability and feasibility: does it solve a real problem for users, does it create meaningful business value, and can it realistically be delivered with the available technology, data and constraints?

That is why the workshop brings together a temporary, cross-functional team of around 6–8 people. The exact roles depend on the organisation, but the team should combine business and process knowledge, people close to the actual work or customer, AI and technical expertise, data expertise and, where relevant, legal or compliance. A decision-maker should also be part of the team to connect choices to business priorities and make decisions when needed.

Having these perspectives in the room at the same time helps prevent technically exciting ideas that solve the wrong problem, valuable ideas that are not feasible, or promising use cases that cannot move forward.

So why should you invest in an AI Opportunity Workshop?

A well-prepared AI Opportunity Workshop requires an investment. You bring six to eight people together, do some preparation beforehand and spend a focused day exploring the problem before deciding what is worth pursuing.

So why not skip that work and simply start building?

Because skipping discovery does not remove the work. It moves the expensive questions further down the road.

You may discover after weeks of development that the problem was not important enough. That the necessary data is not available. That the solution does not fit the actual workflow. Or that people simply do not use it.

The cost is more than the money spent on a pilot. It is also lost time, capacity, management attention and confidence in AI initiatives.

The workshop brings those questions forward, when changing direction is still relatively easy and inexpensive. Some ideas become stronger. Others are stopped early. And new opportunities may emerge from understanding the problem more deeply.

The result is not certainty that an AI use case will succeed. It is something more useful at this stage: better-informed choices about where to invest the next round of time, money and attention.

Frequently asked questions

Where should we start looking for AI use cases in our business?

Start with business outcomes and recurring workflow friction rather than tools. Look for waiting, rework, repetitive information processing, manual handovers and decisions that could be better supported. Then assess whether AI can materially improve the outcome and whether the new way of working can be integrated into existing processes and systems.

What happens in an AI Opportunity Workshop?

In a one-day AI Opportunity Workshop, a cross-functional team of 6–8 people from business and IT moves from existing AI ideas to business goals, real customer or employee problems, workflow context and prioritised AI use cases. The aim is not to generate a long list of ideas. It is to create a shortlist of opportunities with enough business value, feasibility and organisational fit to justify further investment and focused validation.

What do you need to run an AI Opportunity Workshop?

Start with a clearly defined business area or challenge and the business goals you want to support. Bring together 6–8 people who combine knowledge of the business, customers or employees, day-to-day work, technology and data. Existing AI ideas can be useful input, but they are not the starting point for deciding what to build. Finally, reserve a focused day to guide the team from business goals and real problems towards a small number of prioritised AI opportunities.

How do you stop AI pilots from going nowhere?

Start every pilot with a real business problem, a clear baseline, a measurable outcome and an accountable owner. Test the solution in the real workflow with the people who will actually use it.

Most importantly, agree upfront how you will judge whether the pilot is successful. At the end, use what you have learned to make a clear decision: scale, adapt or stop.

How do we calculate AI ROI?

Start by measuring the current cost of the problem. This could be time spent, errors, rework, delays, customer impact or another relevant business metric. That gives you a baseline.

Then compare that baseline with the results of the AI experiment, including the costs of implementation and ongoing operation. Benefits can include time saved, capacity released, higher quality, increased revenue or reduced risk.

Most importantly, agree on what success looks like before you start the experiment. Otherwise, it is very easy to declare a promising pilot a success without knowing whether it actually created enough value to justify further investment.

ready to find which ai opportunities are worth persuing?

Finding valuable AI opportunities does not have to start with a long list of use cases.

A more useful starting point is one business area that matters:

Make the friction visible, understand the people and workflow, and explore where AI could genuinely make a difference.

Every organisation starts from a different place.

Let’s discuss your situation, your current AI ideas and possible next steps.

Book a free 30-minute conversation.

About the author

Jens Broetzmann is the founder of Continno and has more than 25 years of experience helping organisations and teams turn change into practical progress. He combines systems thinking, Agile, Lean, Design Thinking and behavioural change principles to create approaches that fit the context rather than forcing a methodology.

About Jens Broetzmann | Jens Broetzmann on LinkedIn