Build vs. Buy in the Age of AI

TLDR:

AI is making software development faster and more accessible than ever, enabling manufacturers to build applications in days instead of months. While AI can dramatically reduce development effort, it does not eliminate the long-term responsibilities of security, governance, maintenance, scalability, and support that determine whether a solution succeeds over time.

Key takeaways:

• AI has lowered the barrier to software development, enabling manufacturers to rapidly build dashboards, workflows, and single-use applications.
• Building software is only part of the challenge—ongoing requirements like security, governance, integration, support, and maintenance often represent the larger long-term investment.
• AI-generated applications are well-suited for targeted use cases, such as departmental tools, reporting dashboards, and process experiments.
• Mission-critical manufacturing systems require more than code; they demand scalability, reliability, traceability, and long-term support that dedicated platforms are designed to provide.
• The modern build-versus-buy decision is less about development speed and more about ownership, sustainability, and whether maintaining software aligns with an organization’s long-term priorities.

Interested in discussing an upcoming AI initiative? Contact us today.

AI: the death of SaaS?

Historically, manufacturers seeking to implement software and automation required outside developers, new infrastructure, and months or years of effort.

Today, things are different.

With the emergence of generative AI and low-code platforms, it's easier than ever for, well, anyone, to create applications. A motivated engineer, operations leader, or IT professional can describe a problem, generate code, and have a working solution in a matter of days—or even hours. As a result, many businesses have begun asking:

"If AI can help us build software, why would we buy it from someone else?"

It's a fair question, and one that has taken root in the broader marketplace. However. While credence may be lent to this notion, the overarching idea—"why buy when I can build"—falls short of capturing the reality. Building certain types of software has become dramatically easier; owning, maintaining, and updating software has not.

Speed vs. Efficacy

In our industry, digitalization projects are often born of a single objective:

• Track operator tasks
• Manage quality checks
• Collect production data
• Create a dashboard
• Automate a workflow

With today's AI tools, generating such an application can be achieved by a team as small as a single individual. One person with one subscription and a sufficient understanding of their business’s rules, systems, and processes.

On the surface, the value is clear: an AI-generated solution made by an in-house team member circumvents manufacturers’ need to spend time and money working with third-party developers. No contracts, no market research, no RFPs. The decision to build seems without drawbacks.

At least, until questions start to arise:

• Who can access the AI application?
• How are permissions managed?
• How is data secured and backed up?
• What data is the application capturing, and is it complete?
• What if the functionality needs updating?

This truncated list showcases the types of considerations that non-dedicated development personnel (ie, a plant manager) may—understandably—fail to consider when creating an AI-generated application. And so the cracks begin to appear.

Manufacturers are experts at what they do. The same is true for software developers. It would be unrealistic to expect someone from a technology company to drop into a plant and make instant, value-additive changes. And yet, as AI tools continue to grow in power and popularity, the inverse of this scenario is increasingly touted as a practical, viable alternative to dedicated solution development.

Solutioning for the Long Term

One of the biggest changes unfolding across manufacturing is the shift toward standardization. Where once bespoke solutions reigned supreme, more and more companies now seek ease of solution creation, delivery, reliability, security, and repeatability. Homegrown AI applications may fall short in many of those categories.

Another prominent voice in the software conversation? Price. In this way, the advantage seems clear: building a solution with AI is undoubtedly cheaper than working with a dedicated software provider.

At least initially.

When organizations estimate the effort required to build software, they often focus on development, which, as it happens, tends to be one of the smaller costs over the life of the application.

Larger, overlooked costs can include things like:

• Security
• Administration
• Governance
• Backup and recovery
• Infrastructure maintenance
• Integration
• Documentation
• Training
• Testing
• Technical support

As a business’s tech stack evolves, AI-generated solutions will need to be modified and validated. Translation? While AI can fast-track software development, it doesn't eliminate the responsibility, cost, or effort required to maintain it.

Where AI Makes Sense—and Where it Doesn’t

For over thirty years, we have been making mission-critical manufacturing software. Knowing what has gone into getting our platform to where it is today—decades of development, testing, iteration, deployment, and support—do we think an AI application could supplant an MES’s place on the shop floor? No. At least, not in a sustainable, effective, or repeatable way.

Proofs of concept. Projects that target a niche aspect of production. Solving for a small technology gap with a system or asset that operates in relative isolation. In cases like this, the value of AI is immense. It levels the playing field for small and mid-sized manufacturers who—in years past—may have forgone digital optimizations as they did not have the budget to work with dedicated software providers.

These—among other unlisted use cases—are a net positive for our industry.

Build vs. Buy in the Age of AI

AI is undeniably changing the economics of software development. Today, manufacturers can prototype faster, experiment more freely, and solve problems that previously would have required substantial investment.

However. It is worth remembering that, as technology continues to evolve, the need for nuance increases.

A reduction in development effort should not be confused with a reduction in responsibility. Software still needs security. It still needs a roadmap. And the businesses taking on these AI-driven projects? They can still benefit from the insights of dedicated technology providers.

Interested in discussing an upcoming AI initiative? Contact us today.