Five steps to guide your company’s AI journey

But many organizations are facing challenges in this new journey, as integrating generative AI into business processes is rather demanding. This is not due to the complexity of the technology itself, but rather because one needs to identify the perfect fit for its implementation.

The generative AI space is also constantly evolving. Microsoft, Meta, Google, OpenAI and others are continuously developing their Large Language Models (LLMs), creating a daunting mix of technologies, opinions and possibilities.

So how do you ensure your efforts with generative AI deliver real business value, instead of just becoming an expensive and fragmented experiment?

Over the past couple of years, my team has helped a number of customers to map their AI journeys and take their first steps in the space. Here I look at some of the key learnings.

Step 1: Properly organize and label your data

AI and data management go together like hand and glove. It’s simply not possible to have poor data management and expect to develop a successful AI solution. Organizations should thus always ensure that LLMs are using high quality data that is properly labelled and organized.

For example, you need to specify which of your data is of a sensitive nature and which is not. This not only impacts the AI output, it also determines whether your AI solution could be on-premises, in the public cloud or a hybrid model.

Step 2: Identify your use cases

Once your data is in order, the next step is to identify the right AI use cases for your business. Building a backlog of these potential use cases – with defined parameters – allows you to prioritize and focus on those that offer the greatest potential for impact. Use cases should be categorized as simple, medium or complex.

At Tietoevry Tech Services, we work closely with our customers to refine their ideas, create this structured backlog and provide guidance on prioritization. Once this exercise is done, you can cherry-pick which use case to begin with. Normally we advise starting with the low-hanging fruit, i.e. those use cases that are simple to implement but deliver significant business value.

Step 3: Select the right technologies

When the first two steps are in order, then you are in a much better position to choose the right AI technologies for your specific use cases. You have to be very careful and thoughtful in this choice, as the selected technology has a big impact on the success of your AI journey.

Too often, organizations take a technology-first approach, selecting tools before considering the use cases and the business value they aim to achieve. This approach rarely works. Instead, the focus should always be on first identifying the use cases and the business value they bring. Technology should never drive the process – it should enable it.

Step 4: Choose the right project approach

Executing an AI project requires a different approach to that of a standard data project.

In the standard projects that our customers are used to, outcomes are predictable and components behave as expected. The success of an AI project depends on a much wider range of factors, including the quality of the data and the type of model you are using. These and other variables can lead to unpredictable outcomes, making AI projects inherently more complex.

At Tietoevry, we address this complexity through a structured Proof of Concept (POC) or Proof of Value (POV) phase. This allows us to evaluate a case within four to eight weeks and determine whether it delivers the expected business value. Based on the results, we work with customers to decide whether to scale the solution for production, or further refine the use case.

Step 5: Foster an AI-first mindset

With all of the above in order, your company will be in a better position to start thinking about AI from the outset of every project. An AI-first mindset depends on three key elements: processes, people and tools.

The process is about aligning AI strategies with business goals. A clear, well-defined strategy ensures AI initiatives are integrated into the organization’s objectives to drive meaningful results.

It’s essential to provide AI training and keep your teams updated on the latest AI developments. Empowering them with knowledge ensures they can effectively leverage the technology.

Finally, organizations need to make the right tools available. Platforms like GitHub Copilot and other cloud-based AI services enable teams to build custom AI solutions tailored to your company’s needs.