Building an AI Centre of Excellence: what you need to know

By Kelly Gonzalez - Director AI Strategy, Bell Canada, and Jason Milnes – Director AI Products, Bell Canada 

In our recent blog, “Four steps to building a successful enterprise AI strategy," we highlighted the importance of choosing an operating model for AI deployment. Our experience has shown that rapid experimentation with AI can effectively kickstart an organization's AI adoption. However, as AI use cases mature, a degree of stabilization becomes necessary. This can be achieved through a well-defined operating model that balances centralization, governance, and the distribution of AI capabilities, considering factors such as the organization's size, AI maturity, and business objectives.

One effective model is the creation of an AI Center of Excellence (AI CoE). An AI CoE is a dedicated team within an organization that provides AI expertise, accelerates AI implementation, and serves as a central resource for AI knowledge. It drives innovation, standardizes procedures, promotes best practices, and fosters continuous improvement within its domain. It is also key to driving enterprise-wide solutions instead of purpose-built initiatives that focus predominantly on solving specific business unit problems.   

While building an AI CoE can bring significant benefits to your organization, it's not a straightforward task. Too often companies find themselves creating one without achieving true excellence or meeting their initial goals.  

Here’s a closer look at some of the key factors we feel are critical when building an AI Centre of Excellence. 

1. Lay the groundwork upfront 

• • • Define your mission: What's the purpose of your AI CoE? What are you trying to achieve? Are you focused on research and innovation, or are you aiming to deploy AI solutions across your business? Get crystal clear on your mission and make sure everyone's on the same page. 
    • Set goals: Consider goals that are Specific, Measurable, Achievable, Relevant, and Time-bound (SMART). For example, if you're focused on research, your goal may be to file X number of patents within Y timeframe. Alternatively, if you're looking to deploy AI solutions, consider a goal such as increasing revenue by Z percentage within a year.
    • Get leadership buy-in: You need champions at the top to make this happen. Present a compelling case for your AI CoE, outlining the potential benefits and the return on investment. Once you have their support, keep them in the loop on your progress. 
    • Manage change effectively: Introducing AI into your organization is a big deal, and people might be hesitant. Don't try to do too much at once. Start with high-impact projects that address key business needs. Involve your stakeholders early on and be transparent about your plans. 

2. Building the dream team 

• • • Identify your needs: What skills and expertise do you need to make your AI CoE a success? Think about data engineers, machine learning engineers, data scientists, and even AI translators who can bridge the gap between tech and business. 
    • Structure your team: How will your team be organized? Do you need a centralized project management office (PMO) or a more distributed structure, incorporating roles from various business functions such as Security, Legal, HR and Finance? Consider the size and complexity of your organization when structuring your team. 
    • Leverage external expertise: Don't be afraid to leverage the power of external partners. Third-party connections into relevant external groups such as technology firms, academia, start-ups and accelerators, are great contributors to any AI CoE - helping you stay current with industry advancements and best practices.  For example, at Bell, we’ve partnered with Mila - the world’s largest academic research centre in deep learning. Mila researchers work alongside our Bell machine learning and AI teams to apply cutting-edge deep learning neural network techniques to identify opportunities for improving business performance and customer experience.  By collaborating with Mila, our CoE gains access to valuable expertise, enabling us to advance projects and achieve results.

3. Centralized, yet decentralized 

• • • Strike the right balance: Your AI CoE should function as a central hub for AI initiatives, but it shouldn't stifle innovation. Empower your teams to experiment and explore new ideas within a defined framework.
    • Create a sandbox: Set up a development environment where your teams can assess new AI solutions before deploying them across the organization. This will help you avoid costly mistakes and ensure consistency. Teams can conduct Proof of Concepts (PoCs) within this environment, adhering to established standards before transitioning them into deployment. We took this approach at Bell with our Generative AI assistant – Alex. Since its launch, Alex has been a valuable tool - helping team members boost their productivity and creativity.

Building an AI Center of Excellence (CoE) is a strategic investment that demands careful planning and execution. By investing time upfront to define clear mandates and success metrics, identify skill gaps to secure the right expertise, and balance centralization with decentralized innovation, your organization can unlock the full potential of AI.   

We've seen firsthand the success that a well-structured CoE can bring.  Our Data Engineering and AI Center of Excellence has been a game changer for Bell, helping us innovate and modernize across our organization. We are confident that you can achieve the same.

Let's explore if a CoE is the right model for you. Speak with one of our AI experts today.