Build, Not Buy: How AI Changed the Custom Software Question

For most companies, the choice between standard and custom software was never strategic; it was budget-driven. Markus Böhm, Mimacom CEO, explains why AI has broken that economics, and where custom software now delivers an edge that standard platforms can't match.

Key takeaways

• The build-or-buy decision was never truly strategic in most companies. It was economically driven.
• 51% of applications are underutilized and 15% sit entirely idle; the waste averages $18 million per year.
• AI hasn't just made custom software better. It has made it dramatically cheaper: costs and timelines have been cut in half.
• For standard processes, buying remains the right call. For differentiating processes, the calculation has fundamentally shifted.
• There's an additional shift: seamlessly connected custom elements within standard software turn a pure buy decision into a hybrid model, one that changes the perspective on standard software itself.
• The decisive question is no longer whether a company can afford custom software or custom elements, but how it secures that competitive advantage.

One question keeps coming up in my work: build or buy? For the past 20 years, the answer in most cases was standard. The platforms were set, the licenses were running. Then AI hit, turned the market upside down, and gave the question a completely new meaning. Build-or-buy today isn't about what companies can afford anymore. It's about where they want to differentiate from the competition.

Standard software won on price, not quality, security, or feature scope

I've spent a significant part of my career on the standard software side: scaling platforms, closing deals, guiding companies through implementations. What I learned is that most build-or-buy decisions were rarely strategic. They were economic. Sure, feature scope and security played a role too. But on feature scope specifically, companies were paying 100% of the license cost for a utilization rate that rarely topped 30%.

And it makes sense; why would you invest tens of thousands in custom software that does the same thing as an off-the-shelf solution? Building custom software costs money, or at least, that's how it looked before AI: six to twelve months of runtime, teams deep in development, project risk. Standard software was the answer because, for most companies, the alternative simply wasn't economically viable, not because standard was always the better solution.

The result is familiar to many:

• Companies buy platform access and use only a fraction of it
• Processes get adapted to the software, not the other way around
• Consultants get hired for everything that can't be configured

The numbers behind this are unambiguous:

• 51% of applications are underutilized, 15% sit entirely idle
• $18 million in average annual waste from unused software licenses per company
• Only about half of all licensed subscriptions are actively used

When the answer is "buy," companies aren't buying software. They're buying access to a platform and paying for what the majority of users only use a fraction of. The business gets built around a product designed for thousands of use cases, not for one specific company.

So the problem was never the vendor. It was that companies had no economically realistic alternative for their differentiating processes. There's a reason more companies didn't recognize this sooner: the economics simply didn't allow for it.
The decision to go with standard software was, until now, rational, not ideal.

Where custom software has always won

Gartner has long distinguished between "systems of record" (payroll, HR, accounting) and "systems of differentiation". Regarding the former: buy. No sensible CEO built their own payroll system 20 years ago, and none does today.

But for the latter, the systems that set companies apart from their competitors, custom software has always been where the truly successful companies built their edge: industry-specific workflows, customer-centric platforms, operational processes.

Mimacom has been building exactly in this space for over 25 years. The clients who came to us even before AI didn't choose custom software because they had money to spare. They chose it because their competitive advantage was baked into their processes: new business models, excellent customer interactions, additional revenue streams. No standard software can capture that without compromise.

AI hasn't changed software, it's cut the price and the time in half

That's the actual shift, and the debate regularly frames it wrong. AI doesn't necessarily make software smarter. AI makes custom software dramatically cheaper and faster to build. That's a fundamental difference. And it opens up new possibilities: affordable customizations, easy-to-implement scaling, and much more.

• In a controlled study, AI-powered development tools reduced processing time by 55%
• A study of approximately 4,900 developers at Microsoft, Accenture, and another Fortune 100 company shows a 26% increase in the number of tasks completed.
• Around 40% of all newly written code today is AI-generated

With AI, that barrier has structurally disappeared for most companies.

An important caveat is that these gains don't happen automatically. Teams that simply bolt AI onto existing processes see only marginal improvements. Teams that consistently redesign their entire development process capture the full effect. The technology has changed. The discipline of software engineering has not.

Three conclusions I take from conversations with executives

1. The economics have changed, but the strategy hasn't.

When I show the numbers on AI productivity, the reaction is rarely skepticism. Usually it's agreement, followed by a moment of silence. The next question, the harder one, is: where exactly does our competitive advantage lie, and does our current software help or hold that back?

2. Build-or-buy is no longer an either-or decision.

One of our machinery manufacturing clients runs most of its operational processes on proven standard platforms, and at the same time built a custom compliance solution with Mimacom that brings together proprietary supply chain, material, and regulatory data that no standard vendor could ever map. The results are remarkable: 80% of data reconciliation is automated, product series analyses can be carried out in 2.5 hours instead of days, and compliance becomes a strategic advantage instead of a cost factor.

Here's another example. A regulated financial institution didn't want a single piece of data, a single query, or a single internal document leaving for external cloud infrastructure, often the case with standard software vendors. Ten years of institutional knowledge sat unused in PDFs as a result.

Within two months, instead of the originally estimated eight, Mimacom and the client built a fully on-premise AI assistant that makes internal reports queryable in three languages. Answer accuracy rose from 48% to 97%. And with it came no vendor lock-in, zero data exposure, and full control in the hands of the internal team.

3. Those who wait fall behind their competitors

Deloitte has reported that only 34% of companies are genuinely redesigning their business model with AI. The rest are optimizing what already exists. The competitive gap between these two groups will become very visible over the next three years.

Five questions to take stock

I regularly talk with CIOs, product leaders, and executives who consider the build-or-buy decision settled. My take: it's worth questioning again.

• Which processes truly differentiate your company from competitors, and does your current software actually support those processes, or does it constrain them?
• How much of your current software spend goes toward features nobody in your company uses?
• How fast can you respond to a new market requirement, and does your platform slow you down?
• Have you actually recalculated the changed economics of custom development in the last twelve months?
• Do you treat software as an operating cost, or as a strategic asset that reflects how you win?

The companies that asked these questions early are in a different position today, not because they invested more – but because they invested in the right places.  That's the difference between software as a cost factor and software as a competitive advantage.