To Build Or Buy AI Agents? Here Are The Pros, Cons And Complexities

This voice experience is generated by AI. Learn more.This voice experience is generated by AI. Learn more.Steve Karp is Chief Innovation Officer at Unanet with 30+ years of experience in enterprise software, fintech and product innovation.

gettyWhether you build or buy AI agents, be sure the choice is driven by value above all else.

​To build or to buy? Companies have been grappling with some version of that question since the dawn of the digital era.​

During the internet’s adolescence, it meant deciding whether to take a DIY approach to designing and building a business website using in-house resources or to outsource.

Later, with the ascendance of software-as-a-service (SaaS) solutions, the focus of the question shifted to whether to develop custom software internally or to invest in a third-party SaaS offering.​

Now, organizations have access to a growing array of foundational AI models and agent frameworks that give them tools to build their own AI agents to execute tasks, dialogue with employees and oversee other AI agents. The question is whether it’s wiser to build your own agents on top of your existing tech stack or to use software solutions with integrated AI agents.​

As is often the case with fundamental business questions, the answer is, “It depends.”​

As CIO at Unanet, I spend a lot of time evaluating emerging technologies and deciding where it makes sense for us to build, where to partner and where an existing solution can get us to the right outcome faster.

Throughout my career in technology, I’ve learned that these decisions often come down to balancing differentiation and control with cost, complexity and long-term value.​

The case for using AI agents and conversational copilots has become clear: for their ability to autonomously execute tasks; gather and analyze large amounts of information quickly to inform and improve decision making; and put knowledge at people’s fingertips.​

AI agents and copilots will play a growing role in your business if they don’t already. Before you go too far down that road, however, you have to answer the “build or buy” question.

Just because your organization can build its own custom AI platform doesn’t mean it should. Ultimately, it’s a highly nuanced decision that warrants considerable forethought.​ Among the factors to weigh in evaluating the best path forward:​

• Strategic Importance: Is the use case for which the platform is intended truly differentiating? It may not be worth expending extra resources on a tool that isn’t strategically critical.​

• Organizational DNA: Some companies are natural builders, owners and caretakers of technology. Others simply view technology as a means to an end. A company’s culture, heritage, history and DNA with respect to tech should weigh heavily in the build or buy decision.​

• In-House Technical Resources: Do you have people with the right skill sets (software and AI engineering, cybersecurity, data science, compliance, etc.) internally to build AI tools that are secure, governable and scalable? And just as importantly, do you have people to maintain and evolve the tool over time?

• Total Cost Of Ownership: This includes development costs as well as usage, infrastructure, support and maintenance costs. While the subscription price for a SaaS solution with AI agents may look higher than the cost of building a prototype, the costs associated with a built solution can mount over time.

• Integratability: How readily can a built AI tool integrate with the systems of record in your tech stack?

• Governance And Security Around The Tool And Data: How well equipped is your organization to monitor, protect, govern and trace the performance of the tool and the data that feeds it?​

Having established the criteria to guide the build or buy decision, now you can develop a scorecard with the pros and cons associated with each pathway: ​

Topping the list of pros with the build option are control and customizability.

Build may be the best route if a fit-for-purpose tool can provide your business with true competitive differentiation, or it can create value beyond what a purchased solution can provide. That also holds true if there are no existing off-the-shelf solutions to fill a need, support a vital workflow or resolve a glaring issue.​

As for cons, the list starts with the cost and complexity of long-term ownership of a built tool. Costs associated with hosting, databases, integration tools, monitoring, security services, development environments and specialized technical talent add up quickly and may ultimately far surpass the original build cost.

Keeping the tool updated, maintained, secure and in step with the needs of the business is a major commitment. If the tool doesn’t mature along with your business, its functionality and value will diminish rapidly. What if the person or people who were key to developing and maintaining the tool leave the company?

There’s also the matter of fit. Venturing outside your organization’s comfort zone and core business can become a distraction or a drag on the business. What’s more, ensuring the tool meets security and compliance requirements, especially for companies in regulated businesses, can become a huge headache. ​

If the workflows your organization needs to manage and the business problems it needs to solve are fairly universal to your industry or type of business, there’s likely a “buy” solution available to meet your needs.

From the right provider, such a solution comes with a range of advantages. It could be purpose-built for your industry so it tracks closely to your internal workflows, processes and compliance requirements. It should come with governance, integrations, security, support, ongoing improvements and refinements and add-on features.

It should also offer conversational and agentic AI capabilities that work effortlessly within the system and come backed by expert developers who can deliver all this—hopefully with strong insight into your business and industry. Finally, it should come with a greater degree of cost certainty and an overall lower cost than a DIY solution.​

The biggest con on the buy side is that you don’t control exactly what’s in the solution, nor how it will evolve. The provider could also opt to stop supporting a solution, leaving you “orphaned.” Likewise, your business could evolve to the point where the solution no longer meets your needs.​

To build or to buy? For many organizations, the answer may lie somewhere in between, where they buy core systems or platforms, then build DIY solutions on top to augment or complement them.

Whatever pathway you choose, be sure the choice is driven by value above all else.​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?