How CIOs balance build, buy, and blend in the age of AI

Recently, OutSystems hosted roundtables with CIOs from organizations ranging from healthcare systems to financial services, manufacturing to technology companies. What started as discussions about AI strategy also revealed that the build versus buy debate is becoming more complex and more critical than ever as IT leaders juggle speed and agility.

These are not merely theoretical conversations. Technology leaders are grappling with real AI implementations, budget constraints, and real pressure to deliver business value. Their insights challenge much of the conventional wisdom about AI strategy and offer ideas for organizations still trying to figure out their path forward. In this blog, I share the main takeaways from these discussions.

The build vs. buy debate is alive and evolving with AI

Despite predictions that AI would simplify technology decisions, the CIOs who attended the roundtable described the opposite. The traditional tradeoffs they thought they understood have been fundamentally altered by AI capabilities that don't fit neatly into "build" or "buy" categories.

It’s not “either or”: An AI strategy is multidimensional

Several leaders described how the choice between building and buying has evolved into a multi-dimensional decision involving purchased solutions, custom development, and embedded AI capabilities that require entirely different evaluation frameworks. One enterprise architecture leader summarized it as navigating build versus buy versus use embedded AI, creating a three-way conversation that demands completely new evaluation criteria.

At OutSystems, we've seen this complexity firsthand. When we implemented our AI-powered support assistant, we didn't simply "buy" an AI solution or "build" one from scratch. We combined our existing knowledge base infrastructure with AI models, custom prompt engineering, and integration layers to create something that delivered a 38% success rate in self-service resolution. That's not a build-or-buy story; it’s a blend story.

Blend is the way forward

The smart path forward blends both prebuilt and custom software in ways that fit business needs and scale with agility. Leaders at the roundtable were willing to buy AI solutions for commodity capabilities. But for anything strategic or related to competitive differentiation, they were more likely to build. The consensus among these leaders is that AI delivers value when it solves specific problems, not when it provides generic capabilities, which is how many software vendors describe their offerings.

AI specificity delivers real value

Collective skepticism about generic AI solutions was a surprise, given how they’re such a popular topic of conversation. Multiple leaders expressed frustration with vendors promoting AI capabilities that feel more like marketing buzzwords than substantive business solutions.

The CIOs finding real value in AI, especially agentic AI, are those focusing on highly specific use cases with measurable outcomes, particularly in domains where internal teams lack the specialized expertise to build equivalent capabilities. Generic AI promises fell flat, while targeted solutions addressing precise business problems delivered results.

This mirrors our experience at OutSystems. Our community translation project didn't succeed because we used “cutting-edge AI.” Instead, we applied AI to a specific problem (language barriers in developer forums) with clear success metrics. The result was a 65% surge in Japanese developers on our community site within months, followed by expansion to French, Korean, and Portuguese.

In short, the CIOs agreed that AI delivers value when it solves specific problems, not when it provides general capabilities.

Internal ambition meets practical reality in AI development

The gap between AI ambitions and AI realities is a recurring theme for CIOs. There is initial enthusiasm for building comprehensive AI capabilities internally, followed by a sobering realization of the complexity, specialized expertise requirements, and resource constraints involved.

The AI hype cycle compounds this challenge. By the time traditional procurement and development cycles complete, the underlying technology is likely to have evolved significantly. One leader recently told me that building and buying both feel risky because solutions could become obsolete quickly, and the technology changes at light speed.

At OutSystems, we've learned to balance this by focusing our internal AI expertise on areas where we can maintain competitive advantage while using external solutions for commodity AI functions. For example, OutSystems used generative AI to optimize sales processes by analyzing customer call transcripts. We built the integration and workflow orchestration while using proven language models for the core analysis.

Embedded AI drives demand for enterprise orchestration

One of the most significant insights from the roundtable is that embedded AI is a distinct category that requires new architectural thinking. As organizations evaluate their technology stacks, they're discovering AI capabilities built into platforms they already use, including CRM systems, ERP platforms, and specialized industry applications.

This proliferation creates both opportunities and operational challenges. Organizations can use AI capabilities without additional procurement, but managing and orchestrating these distributed AI capabilities becomes a significant challenge.

Several enterprise architecture leaders described the need for what they termed an "enterprise orchestration layer.” It’s a framework for managing AI capabilities across commodity functions they purchase, competitive advantages they build internally, and specialized niche capabilities they source from focused vendors.

This orchestration challenge represents a new category of enterprise architecture requirement that simply didn't exist two years ago. Organizations need platforms that can integrate diverse AI sources while maintaining governance, security, and performance standards.

Clarity on strategic AI investment is needed

Another topic of discussion was how leaders were strategically allocating internal AI expertise versus external solutions. Leaders from organizations undergoing significant transformation described systematic approaches to this challenge. Many create frameworks to categorize applications and capabilities by strategic importance, technical complexity, and alignment with core business competencies. The goal is consistent decisions based on clear criteria that can be updated when technology changes.

Almost all agree that it’s critical to clearly articulate core competencies and competitive differentiators, then align their AI investment strategy accordingly. They buy or use embedded AI for functions that don't provide competitive advantage while investing in custom development for capabilities that directly support their unique value proposition.

Integration becomes the unifying strategic layer

What is clear from these discussions is that future success belongs to organizations that can strategically blend building and buying while maintaining seamless integration across their technology ecosystem.

Several leaders described the operational complexity of managing multiple platforms, diverse development approaches, and varying levels of technical debt. The common theme was using consolidation and standardization to create foundations that support diverse AI implementation approaches while maintaining governance, security, and operational efficiency.

This consolidation challenge isn't unique to AI. At OutSystems, we see organizations struggling with multiple application development platforms, various AI tools, and disconnected systems that create more complexity than value. When they invest in integration and orchestration, they can use best-of-breed solutions while maintaining unified governance and user experience.

Governance determines success more than technology

Several leaders said they believe that the technology challenges of AI implementation, while complex, can be more manageable than governance challenges. Yet, governance is a critical success factor that often gets overlooked in AI strategy conversations. Issues around intellectual property, compliance requirements, and risk management frequently determine project success or failure more than technical considerations.

The leaders expressing confidence in their AI initiatives were those who had established clear frameworks for data access, development responsibilities, and ongoing oversight. The consensus was to treat governance as a foundational requirement. They've learned that AI success depends as much on policy frameworks and organizational processes as it does on model accuracy and technical performance.

The path forward: Embrace strategic complexity

Conversations with CIOs across industries indicate we're moving toward more sophisticated and nuanced approaches to AI strategy. Our own AI implementation experiences at OutSystems bear this out.

Build versus buy is giving way to multi-dimensional strategies that consider technical capabilities, business objectives, resource constraints, and governance requirements.

This complexity isn't a problem to solve. In fact, if your organization can navigate this complexity effectively, you can create sustainable competitive advantage. Those seeking simple answers to complex questions will find themselves falling behind.

As one leader told me, the build versus buy debate will continue to evolve, but organizations now have better tools, better insights, and better frameworks for making these decisions. They can build and buy and blend in ways that serve specific business objectives while maintaining the flexibility to adapt as technology continues advancing.

The build versus buy debate isn't dead; it's evolved into something more sophisticated and more strategically valuable. The question is whether your organization will lead this evolution or be disrupted by it.

Ready to dive into how AI and low-code work together to transform application development? Check out our on-demand webinar, “OutSystems demo: AI-powered low-code for building better apps faster.”