Avatar photo kemtai

Build vs. Buy Part II: Applying the Decision to Computer Vision Technology

We’ve previously explored a decision-making framework to support digital health start-ups in navigating the build vs. buy decision. Here’s how this framework can be applied to computer vision technology specifically.

April 18, 2024  6 min reading

In a previous blog post, we explored the criteria that digital health start-ups should utilize to make effective build vs. buy decisions. These include: (1) cost, (2) time to market, (3) technology quality, (4) maintenance requirements, and (5) connection to core competencies. However, even if criteria are clearly defined, digital health start-up founders can often struggle to apply those criteria to real-world decisions.

So, let’s take a look at the build vs. buy decision-making process for computer vision technology in the digital health and wellness space.

The Role of Computer Vision AI in the Health and Wellness Space

Physical activity has always been and continues to be a critical part of health and wellness. Health care initiatives focused on physical activity and movement, such as the ACSM’s Exercise is Medicine® initiative, are prolific worldwide. Inevitably, a myriad of technologies have emerged in order to support the need to improve health- and wellness-focused exercise practices.

One such technology is computer vision AI. Computer vision AI has emerged as a key technology to support proper and safe exercising because it can analyze individuals’ movements, assess their effectiveness, and, in some cases, correct those movements. It also can arm coaches and clinicians with important data on their patients’ and clients’ performance when they are apart. The applications of this technology vary from physical therapy and rehab to corporate wellness and medical fitness. 


Strong computer vision AI is a very challenging solution to develop and often requires more costs than non-computer vision experts may be aware of. For example, the complexity of algorithms, data acquisition, and model training require substantial investments in expertise, infrastructure, and resources. Additionally, ensuring that the computer vision AI technology is accurate and reliable enough to be used in healthcare and wellness settings adds additional expenses. 

For companies that do not specialize in computer vision, founders often lack an understanding of just how technically complicated ground-up development of computer vision technology truly is. This means that the cost of building computer vision solutions in house is typically very unpredictable, often significantly higher than predicted. This cost underestimation also corresponds with a high opportunity cost for internal development because it requires resources to be pulled away from other revenue generating or cost reducing activities. Of course, not all technical solutions are as costly and unpredictable to develop in-house, but computer vision should not be grouped together with those solutions. 

Time to Market

Just as digital health start-up leaders often underestimate the cost of computer vision, they often downplay the difference in time to market between the build scenario and buy scenario. Developing computer vision AI requires very specific expertise and experience, meaning that even some of the strongest digital health start-up development teams may not be equipped to deliver the technology in a timely manner. Even the leading computer vision AI companies today have required years of refinement and often have teams with decades of experience in computer vision. The digital physical activity, exercise, and rehab markets are not immune to time to market challenges, and founders in the space should understand that licensing computer vision technology provides a clear speed advantage.  

Technology Quality

For some technologies, technology capabilities can be rather binary, meaning that, once implemented, organizations can just “check the box” on whether they have the necessary solution. Unfortunately, that is not the case with computer vision. There is a very wide range of computer vision AI quality. There are many complexities, including the number of body points tracked, the sophistication of the feedback provided, the model’s ability to handle different body types and environments, and more. If an internal development team is not equipped to handle these complexities, the disparity between the quality of their solution and experts’ solution will be very clear.

Maintenance and Enhancement Requirements

Computer vision AI tends to require intensive maintenance and enhancement. First, in digital health and wellness, computer vision AI tends to be a patient- or user-facing technology. This means that keeping the system functioning properly is key for a smooth patient or user experience. Moreover, computer vision AI is simply not a static technology. In recent years, the field has undergone significant technological improvements. For example, only about a decade ago, computer vision technology for body tracking required specific hardware (e.g., special cameras). However, today’s most advanced solutions can run on any device with a camera. Given this pace of change, it is likely that computer vision AI technology will continue to evolve significantly in the coming years. Because of the speed of change in the field and the need for efficient maintenance, relying on an internal team to support ongoing updates to the computer vision solution can be extremely challenging. If a digital health start-up chooses the right computer vision partner, they can be confident that their solution will better withstand the test of time.

Connection to Core Competencies

Digital health start-ups are often focused on providing or supporting clinical care in a more efficient and effective way. Computer vision can play an important role in this, helping to ensure that patients or users are moving and exercising properly. However, computer vision itself is not typically a core competency of these companies. More commonly, digital health start-ups focused on exercise and rehab tend to add most of their value through clinical care or exercise expertise. Devoting significant resources to computer vision technology can pull important resources away from these primary focus areas. Even if computer vision will be a key solution within a digital health start-ups core offering, the benefits of licensing the technology to deliver a superior core product can still make it a worthwhile decision.

What’s the verdict?

Given its development cost and time, the differences in technology quality, and the post-implementation resource requirements, it typically makes sense to license computer vision AI technology rather than develop the solution in-house. Hopefully, the above exercise is a helpful one for organizations trying to navigate the build vs. buy dilemma, whether the decision is for computer vision or some other technology. 

Still not sure what the right solution is and want to learn more? Feel free to reach out to info@kemtai.com. We’d be happy to answer any questions you may have!