Artificial intelligence (AI) is transforming the way work gets done across most, if not all, industries. Looking specifically at medtech product development, how should AI transform how products are developed—and is it doing so already?

One obvious motivation is strictly focused on cost reduction. There is no shortage of fodder in the media about jobs lost to AI, particularly for software engineering. This quickly leads to the conclusion that AI is a cost-saving tool. Not a great story to inspire the innovators of the world to embrace AI, especially when the cost-savings message is often coupled with a reduction in demand for innovators. Thinking about AI as a cost saver fundamentally limits the potential, not to mention the interest in adopting it as an accelerator of innovation.

Microeconomics of Innovation

For innovators in healthcare, I expect most leaders have not experienced a product development program that completed all of the scope that was desired at the time of inception. Early on, development program teams often manage scope tradeoffs to optimize cost and schedule. Concepts like minimally viable product are used as a framework to tightly control scope due to the economic realities of time and/or budget. Businesses, large and small, having to control budgets and time is always a constraint in the competitive medtech market. Finding a mechanism to relax those constraints offers a significant opportunity.

The microeconomics of innovation is about more than just time and budget. It is about engineering innovation capacity. An engineering team that is overburdened by intellectual manual labor, constraints, or inefficiencies more broadly will underproduce. Eliminating these inefficiencies increases the supply of innovation capacity.

One historical example of an organization going to the extreme to eliminate inefficiencies that stifle innovation is Lockheed Martin’s Skunkworks program. Lockheed Martin structured teams to maximize innovation by bypassing common points of friction. This model allowed Skunkworks to achieve what was considered impossible, such as developing the XP-80 Shooting Star in just 143 days.

While this is an extreme example, it does support the concept of reduced friction translating into accelerated innovation. AI represents an opportunity to simulate some of those conditions by automating the tedious tasks that consistently consume bandwidth from talented engineers.

Friction

Friction is any regular activity that is time-consuming, tedious, and ultimately error prone. These points of friction drive inefficiency, consuming a precious supply of engineering capacity. Yet not all points of friction represent a level of dysfunction. Going back to Skunkworks as an example of a low-friction innovation environment, one point of friction was issue resolution between design and production. Co-locating the design and production teams allowed for quick response time to any production issue. Skunkworks involved a lot more than just co-location of the teams, but this example illustrates a point of friction being mitigated through a targeted and relatively simple solution.

Each organization experiences friction at different points in the product development lifecycle. Co-locating design with production was a reasonable solution for Skunkworks, but it is not a likely solution for medtech organizations. But this does represent a pattern to follow. By identifying friction within your product development organization, you are pinpointing the best opportunities for AI to accelerate innovation.

Removing friction from the development lifecycle changes the economics of a program. This, in turn, provides an organization with choices. Expanded scope, shortened timelines, or reduced budget all become possible outcomes within a low-friction innovation environment.

Conclusion

A recent study by MIT found that 95% of AI initiatives are failing to deliver measurable return on investment. A 95% failure rate would suggest AI is a poor investment, but that seems like an unlikely conclusion given the transformative nature of this nascent technology. The question is how should AI be applied to ensure a return?

Lean management principles are a good starting point for an AI initiative. Creating a value stream map of the sequence of actions required to develop a product is a starting point for the identification of friction. Friction leads to “waste” via lost time or unnecessary effort.

The most ambitious or broad AI use case may not be the most valuable, or even achievable. By systematically eliminating friction with relatively simple AI use cases, the product development process evolves naturally. This approach avoids significant disruption by surgically applying AI to your existing processes and, ideally, existing tools. One of the benefits of AI is flexibility—it can be adopted successfully by integrating it in concert with your existing quality system and tools.

Focusing on narrow, high-value use cases is key to success in realizing returns from AI. By systematically eliminating friction, the economics of innovation shift, allowing organizations to deliver more lifesaving or life-sustaining products to patients.