Building an effective AI strategy for medtech organizations

“Do you have an AI strategy?” makes as much sense as asking, “Do we have an Excel strategy?” —Mihnea Moldoveanu, University of Toronto

AI should not be the first thing that comes to mind when we look for new ways to deliver value for patients and providers. Instead, the focus should be on the choices we want to make using AI—a decision-centric approach that yields valuable perspective when evaluating its potential across healthcare sectors.

While AI has recently gained hype in medtech, it’s not new. In fact, the first AI-enabled device was approved in 1995 and over 1,000 such algorithms have since been approved by the U.S. Food and Drug Administration (FDA) as software-as-a-medical device. It’s almost a given that medtech will continue to leverage AI in areas such as diagnostics, remote patient monitoring and robot-assisted surgery. Medtech also will venture into many more novel therapies to drive better outcomes and experiences for patients and healthcare providers (HCP).

There is a significant dichotomy in how AI is being adopted in a typical medtech organization. While there is great innovation happening in “AI in product,” there is much left to be desired regarding AI for internal organizational effectiveness.

Between 2016 and 2023, AI-enabled medical device authorizations grew at a 49% compound annual rate. Yet another survey found only 32% of medical device professionals use AI tools in their day-to-day work. This gap highlights how embedding AI into tangible products has outpaced its integration into core operational workflows within medtech. Organizations must recognize that applying the same innovative thinking to internal operations as they do to product development can create significant competitive advantages through operational efficiency and enhanced decision-making.

Medtech companies can feel frustrated by this divide and don’t know how to start their journey. Common questions include:

• What do we want AI to do to make our organization successful?
• How do we balance the upside with the potential risks of AI?
• What building blocks are required to make the investment worthwhile?
• What role does culture play in adopting AI?
• How do we measure success and business impact?

These questions require significant activation energy to address. As a starting point, we’ve developed a framework that defines the key elements of an AI strategy, enabling medtech organizations to gain a comprehensive view of the critical components they should consider.

A strategy only matters if it creates real-world value. Converting AI plans into quantifiable outcomes requires deliberate measurement and accountability.

The vision and charter. Setting an inspirational and achievable vision for AI that directly ties to business outcomes is the first step. The charter should align what you want to do with AI with what you want to enable for patients, HCPs, institutional customers and employees.

This will be the first test of how honest you are with your AI aspirations. Is it just about getting on the AI bandwagon, or do you genuinely believe you can make a difference?

For example, instead of stating, “We will run an AI project to identify the target physicians who are most likely to churn,” reframe it as, “We will implement a new marketing process to retain customers, using AI to target those clinicians who are most likely to churn.” This prioritizes business objectives over technology and shifts ownership from tech to the business.

With your charter defined, identify use cases across specific domains such as commercial, supply chain, corporate finance and IT. Each use case should link to the value it provides before valuable resources are committed. Focus on business problems with clear ROI potential, such as process and customer experience improvement and revenue generation—not AI for its own sake—to ensure technology decisions are driven by business outcomes and not curiosity.

Prioritization should evaluate both potential impact and implementation effort. Consider factors such as:

• Alignment with strategic objectives
• Potential ROI and timeframe
• Data availability and quality
• Implementation complexity
• Organizational readiness

For our marketing example, you might prioritize the physician churn use case based on its direct revenue impact and availability of robust customer interaction data.

Through KPIs that could be productivity-related, revenue-related or experience-related, it’s vital to have a clear definition of how value, risks, costs and other factors will be measured to justify investments. An ambiguous definition of outcomes and value, as well as the time it takes to see the value, could derail AI efforts and erode trust. For the marketing example, specific KPIs might include reduction in churn rate, improvement in engagement metrics for at-risk physicians and revenue retained from previously at-risk accounts.

Organizational structure and ways of working. Outlining key roles, responsibilities and the reporting structure for your AI organization is essential. This is where the dichotomy mentioned earlier manifests itself most prominently, and it’s often the hardest component to define given organizational culture and existing structures.

The construct can start with three personas:

• AI consumers who will use AI to be effective in their roles
• AI stewards and translators who can bridge domain expertise with data and algorithms
• AI producers who develop the algorithms and applications to be deployed

For an “AI in product” construct, the consumer will be your end customer—the patient or an HCP. For all other use cases, your AI consumers will be internal. Your organizational structure should bring together these three personas to interact and work toward common goals. Everything else—how centralized, regional, divisional and functional—are important questions that can be resolved through effective change management if the foundational principles of the three personas are addressed and purpose alignment exists.

Governance and risk management. If not approached carefully, AI will continue to be presented as a shiny object and panacea for all problems. The benefits of AI must be balanced with sufficient rigor to mitigate ethical, privacy and security risks. Understanding the potential impact associated with AI risks is critical. The term governance is often associated with bureaucracy and slowness, but that should not be the case. Good governance and stewardship foster healthy dissent and dialogue that counters AI exuberance with appropriately conservative approaches.

Regulatory compliance remains a significant hurdle, with the FDA’s January 2025 draft guidance addressing the growing regulatory demands for developers of AI-enabled medical devices. These challenges highlight the importance of robust governance frameworks that address the unique aspects of AI validation and documentation. Successful organizations proactively incorporate regulatory considerations into their AI development processes from the beginning, creating transparent documentation practices and validation protocols that anticipate evolving regulatory requirements rather than reacting to them after implementation.

Critical enablers that support your AI infrastructure. The enablers are the building blocks necessary to develop an AI capability. Without them, AI will remain a pipe dream.

Data strategy. Data is the cornerstone of any successful AI implementation. Without high-quality, well-governed data, even the most sophisticated AI technologies will fail to deliver value. Organizations with mature data strategies are more likely to achieve ROI on their AI investments than those without. Clean, accessible and comprehensive data isn’t just a technical requirement. It’s a strategic business asset that enables everything from accurate predictive modeling to reliable gen AI applications. No data strategy means no effective AI strategy.

A comprehensive data strategy is a long-term plan that defines the people, processes and technologies needed to create, process and use data to intentionally drive value through AI in a meaningful, secure and transparent way.

Implementing this strategy requires addressing several key dimensions:

• Data governance. Establishing clear ownership, quality standards and access protocols that balance security with accessibility
• Data architecture. Designing systems that connect disparate data sources while maintaining data integrity across the organization
• Data quality management. Implementing processes to ensure data is accurate, complete and up-to-date
• Data literacy. Building organizational capabilities to understand, interpret and effectively use data

A sound data strategy enables training your AI in a hygienic, comprehensive manner with minimal biases and maximizes the value of data across use cases. Organizations that prioritize their data foundation before racing to implement AI solutions see significantly higher success rates and faster time-to-value. For the marketing example, this might include integrating data from CRM systems, sales interactions, product usage and customer support into a unified customer view that enables accurate churn prediction.

Technology foundation. This encompasses the AI technologies, platforms and tools that will drive capability. Investing in scalable, future-proof technologies that allow for interoperability is key. For technologies that AI consumers will interact with, a rich user experience is important. Investments in these tools and platforms should be purposeful and aligned with business goals.

A comprehensive technology foundation should balance gen AI and classical AI approaches. While generative AI excels at natural language processing, content creation and unstructured data analysis, classical AI methods remain essential for structured data analysis, rule-based decision-making and statistical modeling. Organizations achieving the greatest impact are pairing these complementary technologies—using generative AI to enhance user experiences and knowledge management while leveraging classical AI’s strengths in data-driven prediction and optimization. This balanced approach ensures that each AI technology is applied where it delivers maximum value.

Given rapid changes in the technology landscape, closely aligning with key platforms that have demonstrated agility is essential.

Talent and skills. The competition for AI talent will be intense and challenging. Having a solid talent strategy that includes hiring new talent, upskilling existing talent or partnering with external experts is crucial.

The healthcare AI talent landscape presents significant challenges that demand creative approaches to building AI capabilities through strategic hiring, partnerships and upskilling programs. Forward-thinking medtech organizations are responding by creating hybrid teams that pair healthcare domain experts with AI technical specialists, fostering cross-disciplinary collaboration that bridges the knowledge gap while building internal capabilities organically.

While in some cases AI will replace human talent, in most cases, it will augment human capabilities, leading to the emergence of new roles at the intersection of AI and other disciplines such as anthropology, ethics, policy, arts and engineering.

Partnerships and ecosystems. Avoid an insular approach to AI development. While organic development is necessary, leveraging technology partnerships with hyperscalers, chipmakers and service firms—as well as with academic institutions, think tanks and your customers—will become increasingly important for credibility. Moreover, being a partner who actively contributes to the healthcare ecosystem of HCPs, payers, patients and fellow manufacturers is vital to realizing and capturing value through AI.

Value realization and business impact

After defining use cases, assessing risks and deploying enablers, the next step is to build and deploy a mechanism to continually monitor whether value is being realized and captured. Monitor leading and lagging indicators that help understand the value for your AI consumers in terms of stakeholder voice, productivity, revenues and other metrics. Conduct periodic reviews where robust discussions occur on value realization, business impact and stakeholder satisfaction to lend discipline to the rationale behind ongoing AI investments.

A balanced scorecard that effectively ties all AI initiatives together with their impact will facilitate these discussions. For our marketing example, this might include tracking churn reduction rates, engagement improvements and actual revenue retention compared to projections.

Change management and adoption. Ultimately, it’s about the people. We are here to serve people through people and AI is just one of the tools to ensure that this happens. Ensuring that stakeholders view AI capability as a means for them to achieve their personal and professional goals is important. Purpose and culture trump any initiative, so continual messaging that is trust-based—emphasizing that these AI initiatives and use cases are only a means to an end—is critical. Only then is there any chance of reconciling organizational purpose with what you want to achieve with AI.

The intent behind this article is to ensure we see the elephant as a whole and not just its individual parts. By establishing a comprehensive AI strategy that considers all these elements, medtech organizations can strike the right balance between innovation and practical implementation.

Start by evaluating where your organization stands on each component of this framework, identify the most significant gaps and develop a roadmap to address them systematically. Remember that AI strategy is not about the technology itself but about how it enables better decision-making and outcomes for your business and customers.