Bridging AI and Business Requires Leadership

Four out of five AI pilot projects fail. The common challenges stem from insufficient business guidance, a lack of expertise, and poor data quality. As the number, scale, and criticality of AI projects grow, it’s wise to clearly assign responsibility for managing these aspects.

Adopting AI shouldn’t be solely IT’s responsibility. It’s a comprehensive and long transformation that affects the entire business, requiring every unit and team to evaluate how their workflows, competitive landscape, and products will evolve with the support of machine learning tools.

As this transformation inevitably spreads across the organization, it becomes increasingly difficult for IT to manage alone. Our recommendation is that any significant AI adoption should be guided by an AI Center of Excellence—or simply a dedicated AI hub—a virtual team specifically tasked with steering this change.

An AI Center Minimizes the Cost of Missteps

Piloting AI—especially language model-based solutions—is incredibly easy. All it takes is an enthusiastic developer—or even less if using low-code tools or Microsoft Copilot. However, challenges emerge when there’s a desire to scale the piloted solution toward production.

It’s perfectly acceptable for a pilot to fail—that’s the essence of an experimental culture. However, a pilot should test precisely the problem it was designed to address. For example, an AI pilot may rightly conclude that AI isn’t the solution to a particular business problem, but it shouldn’t fail due to IT or administrative reasons.

The role of an AI Center of Excellence is to support projects from planning to production and through to maintenance, ensuring that missteps are minimized. Here’s how:

• Support business in identifying the right use cases: It’s crucial to carefully select the areas where AI is implemented. Experience and expertise play a key role in making the right choices.
  
  
• Bring expertise into the planning phase: During the project planning stage, the AI Center of Excellence contributes by leveraging insights from previous AI projects, industry-wide experiences, and identified risks. This support helps pinpoint the most likely risk factors for the pilot and allocate resources accordingly.
  
  
• Provide support during project execution: The AI Center offers implementation support, such as AI-focused technical assistance and project management. When is the right time to test the quality of an AI model, and how should it be done? If the quality isn’t sufficient or model performance declines, what can be done? How should AI-related risks be managed?
  
  
• Ensure functionality and compliance during production deployment: The EU AI Act and other regulatory frameworks require AI solutions to be classified, documented, and governed. The AI Center takes responsibility for this centrally, allowing other units to focus on developing the business.
  
  
• Offer support services during the maintenance phase: Monitoring AI solution usage, measuring the resulting business value, cost management, model updates, and more require both technical and logical expertise. Maintaining these capabilities for each individual project is expensive. The AI Center also keeps an eye on market developments, ensuring existing AI solutions remain efficient.

Who Should Be a Part of the AI Center of Excellence?

In most organizations, an AI Center of Excellence is best established as a virtual team. Representatives from across the organization are selected to support AI initiatives on a part-time basis. Full-time members become necessary when development functions are centralized, or the volume of projects increases significantly. Often, AI centers also include external consultants, as maintaining sufficient expertise—especially in a rapidly evolving technical domain—can be challenging for many companies.

Ideal candidates for the AI center include the following. The leader should be a business-oriented individual with a natural interest in AI solutions. Members should bring expertise in IT (system reliability, AI models, and cloud services) and legal (compliance and regulations). It’s also recommended to involve representatives from HR, finance, and communications. Highlighting even small AI successes across the organization and conducting mature business calculations on the value created helps ensure continuity. Additionally, it’s essential to consider the impact on personnel, from skill requirements to role transformations.

A diverse AI center may begin as a discussion forum but can evolve into a formal organizational body with a critical mission. The center defines the organization’s everyday AI strategy: How should AI be used in our company? What are our goals, how will we evolve in the coming years—and how do we prepare for that as individuals?

If you need support in organizing your company’s AI initiatives, get in touch. We assist businesses of all sizes in identifying and implementing business-critical AI solutions while also creating sustainable and efficient governance models for them.