Build versus Buy for Pharmaceutical AI: Should I Build or Should I Buy?

Most pharmaceutical companies have long since accepted that AI is and will continue to be essential to maintaining a competitive advantage. The big question facing pharma today, when it comes to AI, is whether to Build or Buy. That is, whether to build a bespoke AI solution in-house or buy, either as a subscription or a one-time purchase, an existing, “Commercial over the Counter” (CotC) solution.

Every pharma company is different, and business leaders will need to think carefully about their needs, business objectives, and unique strengths and weaknesses. In this article, I outline some of the main pros and cons of building vs. buying AI for pharma and offer some key questions for pharma business leaders to help in making a decision.

Pros of Building, Cons of Buying

While building is usually a more expensive and labour-intensive process, there are definite advantages to consider in comparison with buying.

Customization. If the functionality of a CotC solution fails to map onto your business needs, it can be a bit like trying to fit a square peg into a round hole. Complicated work-arounds come into play, eroding the cost benefits of buying. Increased complexity can also lead to frustration and resistance from key stakeholders. Building in-house ensures you have everything you need, nothing you don’t.

Flexibility. Even highly customizable CotC solutions are driven by market forces. Updates and new features cater to the majority, making it difficult to grow your competitive advantage. Building gives you full control over code, so innovative, unique solutions can be added on the fly.

Ownership. There are two things to consider here. First, full ownership of the system itself means you can modify it as you see fit, patent innovative solutions to protect your competitive advantage, and even market your system as a proprietary solution down the road.

Ownership of data is also an important consideration. While CotC providers are unlikely to outright steal your data (and those in the pharma and healthcare industry must abide by HIPAA and other regulatory requirements), your company and customer data will still contribute to the provider’s machine learning (ML) engine. You may be feeding a potential future competitor with valuable data.

Vendor Lock-in. Vendor lock-in can take a variety of different forms, from the explicit (e.g., a signed contract stating a minimum commitment) to the practical (e.g., the difficulty of decoupling an already integrated AI ). In addition, you may find yourself at a critical disadvantage if your vendor of choice doesn’t keep up with its competitors, or fails to implement some new technology or feature that would greatly benefit your business.

Pros of Buying, Cons of Building

Quick returns on investment. Building is typically a long, complex, and expensive process. It requires hiring data scientists and developers, extensive research regarding regulatory compliance in multiple regions (typically built into CotC solutions), and, inevitably, unforeseen expenses and delays. CotC solutions can typically be integrated into existing workflows within a matter of weeks, whereas building can take multiple quarters.

On-demand experts and support. CotC providers have entire teams of accomplished, expert developers dedicated to helping clients—and especially large clients like pharma and healthcare companies—to quickly resolve issues. Of course, an in-house team can meet this need as well, but requires huge investments in hiring and training. While you have less control over incident resolution with CotC solutions, you know the team on the other end of the phone is well-equipped to deal with anything.

Training data. To get the most out of machine learning, it needs to be trained on data—lots of data. Many pharma and healthcare companies do indeed have extremely large datasets to train their own in-house AI, but CotC solutions will have been trained on data sets many times or even orders of magnitude larger.

On the flip side, if you hope to you train an AI on your data and yours alone, this may be difficult with a CotC, which is typically designed to learn from the collective data of all its corporate users.

Collective experience & knowledge sharing. Machines aren’t the only ones who benefit from larger data sets. Human experience and knowledge is a precious resource. Going with a CotC solution means benefiting from the provider’s experience of working with dozens, if not hundreds or thousands, of other companies whose needs closely match your own.

Cost. It is almost always less expensive to pay for an existing CotC AI solution than to build your own. CotC providers will have invested millions or even billions of dollars on their AI solutions, and can recuperate this by selling access to it. But you will need to ensure that you can recuperate the very significant initial and ongoing investments of building in-house, if you decide to take that route.

Key questions for Pharma Business Leaders

Think critically about the following questions when deciding whether to buy or build.

Will the new system be a core feature or critical function of your offering?

If so, it’s likely better to build. You’ll want the developmental control, ownership over data and models, and flexibility to grow and evolve that comes with building. If not—if you’re merely looking for AI solutions to, for example, empower your marketing team or squeeze more value out of existing data—a CotC solution will offer quicker returns.

Will CotC providers offer a unique competitive advantage?

If you are able to use a CotC solution in an innovative and unique fashion, one that will help you stand out from your competitors, then do so. There’s no need to spend the time and money on a bespoke solution in that case. But by virtue of the fact that your competitors may already be using the same solution, you may not get the edge you were after.

How greatly do CotC solutions differ from your needs? What would be involved in fitting it into your existing business model and workflows?

If the perfect solution already exists, there’s little reason to reinvent the wheel, unless you have specific concerns over data ownership and use or flexibility for known upcoming changes to business processes. If, on the other hand, your business needs are significantly different from existing solutions and fitting them into your business would require extensive, costly, complicated work-arounds, you may be better off building.

Is flexibility essential to business outcomes? Is your market changing rapidly?

This could almost be considered a rhetorical question for pharma, where markets are changing rapidly and disruption is common. If the desired outcomes of implementing an AI solution are liable to change rapidly in the future, you’ll likely find an in-house solution can be updated and modified much more quickly than large providers can manage.

Middle ground: Working with AI specialists

“Build or Buy” is an important question, but the suggested dichotomy is a bit reductionist. There is middle ground. Working with a specialist company can help you decide whether to build or buy, of course, but may also be able to offer alternative solutions—like making use of a CotC for one component of an otherwise bespoke system.

A variety of sophisticated, cloud-based and on-premises “ready-made environments” also exist now, with features and tools that can be used to build very unique, highly customizable solutions. Open-source tools, like TensorFlow by Google and Cognitive Toolkit by Microsoft, reduce the costs and complexities of building in-house, while providers like AWS offer a host of AI-powered building-blocks, such as its Deep Learning neural network functionality and Comprehend natural language processor (NLP).

Alternatively, large companies can build highly productive and unique partnerships with CotC providers, giving them greater control over many of the buying “cons” listed above. Doing so requires an intimate knowledge of the landscape, which is where an experienced consultant can provide invaluable advice.