Designing Enterprise AI Strategy: Key Pillars for Maturity and Transformation
By Tze C Chiam, PhD, Senior Director, Health Systems Analytics and Modeling, Connecticut Children’s | Assistant Professor, Department of Pediatrics, UConn School of Medicine
Artificial Intelligence (AI) and Machine Learning (ML) solutions have been used in various industries for decades. However, the emergence of Large Language Models (LLMs) such as ChatGPT in the recent years has dramatically increased public interest and organizational focus on AI/ML applications. As a result, many companies are now taking deliberate steps to accelerate their AI maturity journey. Like any transformative initiative, success demands a clear strategic and operational framework, strong organizational alignment, and sustained leadership support. This article serves as a guide for organizations aiming to develop or refine their AI/ML strategies. The following components are essential for building a successful, enterprise-wide AI transformation:
Developing and executing an AI/ML strategy is a complex yet rewarding endeavor. Success requires more than deploying technical models—it demands leadership, governance, talent, infrastructure, security, and a culture of learning and collaboration.
- Executive support and leadership commitment: Organizational transformation must be championed from the top. AI/ML strategy cannot be treated as a pet project. Instead, it requires committed executive sponsorship to ensure alignment with the organization’s broader strategic goals. As emphasized by [Iansiti and Lakhani, 2023], “…it is critical for leaders to be committed to leading the transformation, to be all in”. Executives should also be willing to challenge the status quo and recognize the architectural shift needed to provide a sustained commitment to transformation.
- Support Structure and Organizational Design: Organizations typically adopt one of 4 analytics support models or a hybrid of: centralized, decentralized, federated, or hub-and-spoke. Each structure has advantages and trade-offs. The optimal structure depends on factors such as the organization’s size, complexity, culture, available analytics resources, and strategic objectives. Similar considerations apply to creating an AI support structure for each organization. AI/ML teams should be structured accordingly to balance autonomy, governance, growth, and resource efficiency.
- Talent and skills: With the rapid evolution of AI/ML technologies, organizations must invest in developing talent with both technical and interpersonal capabilities. Ideal data scientists should possess the curiosity and aptitude to keep up with AI/ML technology development, as well as technical skills in data analysis, probability, machine learning, statistics, deep learning, and artificial neural networks. They should also have proficiency in programming languages (e.g., Python, R) and be skilled in data visualization. Data scientists should also be trained to identify biases in datasets used for model training and develop approaches to mitigate these biases. They should be experts in the model training, validation, testing, and monitoring framework to ensure models are developed and deployed through proper due diligence. As the role of data scientists has evolved and analytics is a team sport, data scientists should also have the communication skills to explain complex AI concepts to nontechnical stakeholders.
- The AI factory and technology infrastructure: To enable data scientists to work efficiently, institutions should invest in a modern “AI factory”: A robust, scalable technology stack that supports data pipelines, algorithm development, software infrastructure, storage and analytics tools, experimentation and deployment tools [Iansiti and Lakhani, 2023]. Companies like Netflix have leveraged this approach to drive innovation on a large scale. A strong infrastructure accelerates development cycles and improves reproducibility, collaboration and security.
- Partnerships and Innovation: Successful AI adoption often hinges on collaboration, both internal and external. Internal collaboration across AI technical experts, business stakeholders, and Information Systems professionals is crucial for aligning AI efforts with business needs. To facilitate such alignment, the cross-functional team can also consider having shared performance goals. Externally, strategic partnerships with innovators, hyperscalers, cloud providers, and research institutions can provide organizations access to cutting-edge technology and scalable tools.
- Information Security: Protecting sensitive data is paramount to any organization. This is especially true in regulated industries such as healthcare and finance, where data consists of sensitive information about individuals. Organizations must implement stringent data governance, access controls, and cybersecurity protocols. As cyber threats become more sophisticated, continuous investment in security infrastructure and employee training is necessary to safeguard AI initiatives.
- Learning Community: For AI/ML to gain traction across the enterprise, increasing organizational AI/ML literacy is essential. Creating learning communities fosters a culture of continuous education, experimentation and engagement. These communities can focus on knowledge sharing and innovative ideas generation, technical training and skill development, project-based support and mentorship, or cross-functional networking. Such initiatives can accelerate adoption and ensure alignment across teams.
- Ethics, policy and governance: As of this writing, 31 states in the United States have adopted resolutions or enacted AI legislation [nscl.org]. Regardless of legal requirements, companies that wish to utilize AI/ML technologies should proactively ensure ethical and responsible use of AI/ML technologies through company-wide policies. Establishing an AI governance committee to promote transparency, fairness, and accountability is critical. This committee should also guide compliance, risk management, and stakeholder communication.
- AI/ML Inventory and Risk Management: Maintaining a centralized AI/ML inventory of work and assets is crucial to the organization in multiple ways. It enables organizations to track and manage the utilization of their AI assets in order to inform future investments. An inventory also allows organizations to assess potential risks such as data privacy, security vulnerabilities and compliance issues associated with each tool. As discussed above, although not every state has formal AI legislation, some states, such as California and New Mexico, have enacted legislation requiring state agencies to inventory their AI systems. While this requirement has not been extended to private companies currently, adopting this practice proactively prepares organizations for future regulation and improves internal oversight.
- Return-On-Investment (ROI): Implementing an AI/ML strategy is a significant investment for organizations. Measuring its impact is crucial to ensuring long-term support and prioritization. ROI can be quantified through operational efficiencies, employee productivity, customer/patient satisfaction, revenue growth, cost savings, quality of patient care and adverse events prevented. There are, however, challenges in quantifying such ROIs, including poor data quality, the complexity of identifying causal metrics, and any type of delayed benefits (such as delays in accounts receivable). In some cases, culture shifts or innovation may provide intangible yet valuable returns that are challenging to quantify.
Developing and executing an AI/ML strategy is a complex yet rewarding endeavor. Success requires more than deploying technical models—it demands leadership, governance, talent, infrastructure, security, and a culture of learning and collaboration. By investing in these foundational elements, organizations can accelerate their AI maturity and unlock transformative value in a responsible and sustainable way.
References:
Iansiti, M and Lakhani, K (2023). Competing in the Age of AI. Harvard Business School Executive Education. Harvard University.
National Conference of State Legislatures. Artificial Intelligence 2024 Legislation (2024). https://www.ncsl.org/technology-and-communication/artificial-intelligence-2024-legislation
Acknowledgements:
Advancing AI maturity is a team sport. I’d like to acknowledge the following colleagues who are part of the AI leadership team with me at Connecticut Children’s who have been my thought partners on this journey:
Jonathan Carroll, MBA, Chief Information Officer
Richelle deMayo, MD, Chief Medical Information Officer
Deb Pappas, MBA, Chief Marketing and Communications Officer
Lori Pelletier, PhD, MBA, Chief Quality and Safety Officer