From AI licenses to business results: Why the process comes first

“We tried AI, but it didn’t deliver.”

I hear this in many organizations. Licenses have been purchased, teams have access to large language models (LLMs) such as Claude, ChatGPT, or Copilot, yet the results are not there.

The return on these investments is a fair question, and in many cases it is still difficult to answer. Often, companies simply make the tools available and tell people to use them. It is like buying a car, filling the tank, and asking people to drive, and then asking them later where the productivity gains are.

Artificial intelligence (AI) will change how we do business and how we work. But its value depends on using it in a structured way to solve real problems. Used without direction, it rarely brings measurable results.

More tools do not guarantee more results

When AI is not linked to a specific business problem, people mostly just explore it. Employees test the tools, see the potential and use them to make their daily tasks easier. This has some value, but it does not necessarily mean higher productivity or a solution to a real business problem.

The most common mistake I see is bringing in AI without first improving the process. Our approach is clear: organizations should improve their processes and use AI to support that improvement. Using AI on a process that has not been rethought only automates waste. It makes activities faster that are inefficient or badly designed.

So the starting point should be the process itself. How should the work be done? Which steps can be removed? What should be done differently? Only then should AI be applied, where it adds value.

Where the time saved actually goes

The answer depends on how the organization uses AI.

When companies make the technology available with no further guidance, the time saved stays with each employee. Their work becomes easier and some tasks take less time, but the impact on the business is limited.

When companies follow a structured approach and rethink their processes and ways of working, the time saved reaches the client. Processes become faster and clients get better outcomes.

So this is a strategic decision. Organizations can use AI to improve their processes and serve their clients better, or they can use it without direction and let its benefits get lost across the company.

Where to begin: The client promise

If the process is the problem, the first step is to look at the organization and its teams and ask some basic questions. What are we delivering to our clients? What is our mission? What promise have we made to them?

With those answers, each current process should be reviewed against that promise. Does every step contribute to it? Does it add value for the client? Steps that do not should be removed, and the process redesigned.

In practice, this means:

• Analyzing and mapping current activities.
• Asking whether each of them makes sense.
• Redesigning the process.
• Using AI tools to support the new process and improve what is delivered to clients.

The final goal is to improve the value delivered to clients and, as a result, get good results for the company.

From process redesign to an AI roadmap

Once the process has been rethought, the next step is to define an implementation roadmap. This roadmap combines several initiatives: some focus on redesigning the process itself, and others are related to AI.

AI initiatives usually fall into two categories:

Table 1 – AI initiatives: Business AI and Transformation

Business AI is a fast way to get started. By training people to use large language models such as Claude, organizations can speed up parts of their processes without starting a large-scale project. What matters is that these improvements stay connected to the bigger transformation. Each change to a task should be part of a process redesign with a clear goal.

Results in practice

We are currently using this approach with several organizations. Two cases show different levels of complexity.

Hospitality

At a hotel group, some client reservations went automatically into the property management system, Opera. Others arrived by email and had to be handled manually. We set up an LLM-based agent that reads these emails, takes out the relevant information and sends it into the system through automatic integrations. This was the more structured of the two projects. It needed more expertise and combined AI with coding and system integration.

Insurance

At an insurer, a team leader had to sequence the team’s tasks manually. Using Claude, we built a simple HTML-based tool that does this sequencing automatically. The team leader now saves three hours a day and uses that time to plan and improve the work with the team. In this case, a language model was enough to create a practical solution with very little IT work.

These cases have two things in common:

• They started with a business need: Each project was designed to solve a real problem, improve a process, and deliver more value to the client.
• They were built with the people who use them: Each solution was developed together with the business teams, which helps people really use it and gives a solid base for the future.

Deciding which processes to automate

The first question is whether the process is needed at all, and whether it still makes sense to keep doing it.

The second is how much time it takes. A task that takes one hour a year is probably not the right place to start.

In short, two criteria matter: relevance and impact. Every initiative should solve a real problem and deliver a measurable result, rather than follow a trend.

The first steps for companies that already have licenses

Organizations that have already invested in licenses should focus on two things. The first is practical, hands-on training, so that people can use the tools and understand their potential. The second is building detailed use cases that show what can be done.

Daily huddles are a good example. Many teams already follow their KPIs in Power BI dashboards. Using AI to analyze those KPIs, give insights, and prepare the daily huddle is a simple first use case that makes these meetings more effective and more efficient.

A final reflection

If there is one idea I would like to leave with leaders, it is this: AI should be driven by purpose, never by trends or impulse. The real opportunity is in rethinking the way we work.

Adding these tools to the same processes and the same ways of working will not lead to different results. A useful exercise is to imagine building the company from scratch and ask how its processes would be designed.

With that clarity, AI can support a new way of delivering value: a better customer experience, more satisfied clients, shorter lead times and better results. That is where its real impact is.

At Kaizen Institute, we help organizations turn AI investments into measurable results. We combine our process improvement expertise with Business AI training and transformation projects, always starting from the client promise and the real problem to solve. If your organization already has the tools but not yet the results, this is a good place to start the conversation.