It’s a question leaders across the healthcare ecosystem are grappling with, so it’s no surprise it’s a question I’m getting asked with increased frequency: Does it make more sense to build AI solutions in-house, or buy them from a vendor?

Before you assume my answer because of my role at Infinitus, I’ll be the first to say that some healthcare enterprises are building their own AI solutions today and the decision is paying dividends. Not every organization should buy, and not every organization should build. 

But for most decisionmakers, the right answer may not be immediately clear. Below, I’ve broken down how I view the decision, and have shared the counsel I give when I’m asked this important question. 

The case for building your own AI solutions:

For the vast majority of healthcare leaders, the question of whether to buy or to build is only about the AI agent solution itself – very few will also build the orchestration layer necessary for AI agents to function, and virtually none will build the elements necessary for the foundational layer. 

But there are some things healthcare orgs that choose to build will have in common. Building may make sense if:

1. You’re a very well-resourced enterprise with true software DNA

I can think of two examples of large payors that have proven to be very capable of building and maintaining their own AI. Their organizations have large, specialized engineering teams, internal ML infrastructure, and perhaps most critically, the ability to maintain agent systems over time (I’ll talk more about this below).

2. You have a small number of very constrained, very specialized, and repetitive use cases

In the rare case that you have only one or two use cases and are confident you won’t expand beyond them, it’s possible that building in-house might make more sense than buying. This is also the case if those scenarios have extremely specific needs – and, for example, it may take more work to educate a vendor on those needs than to execute on them yourself. 

3. You want to own the IP or create differentiation in your core business

Building your own AI agent solution can make sense when the capability itself is core to your organization’s competitive differentiation, meaning the logic, data, or workflows the agent handles are so central to your value proposition that you need full ownership, customization, and control. 

The case for buying AI solutions:

Ultimately, most organizations end up buying solutions – even those that start down the path of building their own. Once healthcare leaders encounter the true operational burden of maintaining AI systems at scale, it becomes clear that buying AI solutions makes sense in most instances.

Here’s why:

1. You’re facing complexity that far exceeds what internal teams can maintain

While it’s easy to spin up an AI agent demo, the real work begins after deployment. Large healthcare enterprises often have hundreds or thousands of workflows, each of which requires ongoing monitoring, updates, drift detection, and consistent performance across agents. Most teams simply aren’t built to run this level of AI infrastructure long-term. 

2. Healthcare orgs are healthcare first – not technology first

And that’s a good thing. But it’s historically why healthcare has depended on partners to help them deploy and maintain complex systems. Only a tiny subset of organizations have the engineering muscle to build and support enterprise-grade AI themselves. For everyone else, buying aligns with both capability and precedent. 

3. Dedicated AI partners probably do it better

Companies focused exclusively on AI agents will advance faster than any single enterprise team can – that’s just a fact. Their entire business is building (and continuously improving) specialized agent architectures, safety layers, evaluation frameworks, and so forth. Buying allows healthcare organizations to benefit from that compounding expertise rather than trying to recreate it internally. 

The questions healthcare leaders should ask next

Once you understand where your organization sits in the build-vs-buy framework, the next step is asking the right questions: Which workflows create the most friction today? Where could AI meaningfully change patient or provider experience? Which parts of your operational ecosystem require flexibility, and which require long-term stability? 

The organizations that move confidently into the AI era will be those framing AI in the context of their operational goals, regulatory constraints, and patient impact. Build vs. buy is only the beginning of that evaluation. The real work is designing an AI strategy aligned to what your organization actually needs to deliver.

And of course, I get asked plenty of questions about AI strategy, as well. It’s something we covered in a recent webinar on the questions to ask when evaluating AI solutions. You can also connect with me on LinkedIn – I’d be more than happy to discuss.