Build or buy AI: Rethinking the conventional wisdom
Written by Jens Eriksvik & Simon Lanngren
AI is transforming industries, but many businesses approach it with outdated assumptions. The "build vs. buy" debate oversimplifies a complex decision. Instead of choosing between in-house development and off-the-shelf solutions, businesses should rethink their entire approach to AI - focusing on long-term adaptability, the true cost of ownership, and where they should not invest.
Short on time?
Read the condensed summary to get a quick introduction to the topic.
Assessing business needs
Most organizations overestimate their need for custom AI solutions. Many proprietary systems are underused or become obsolete faster than anticipated. If your business doesn’t have the data, talent, or willingness to make continuous improvements, building your own AI might be a strategic mistake.
On the flip side, relying entirely on pre-built tools often locks businesses into rigid frameworks. These tools are optimized for broad use cases, not your specific needs. If differentiation is part of your strategy, the ease of pre-built AI might ultimately limit your potential.
Unless your company is ready to commit to AI as a long-term capability, your best strategy might be neither building nor buying but simply staying out. Given AI’s capabilities, and what others are doing, that, however, would seem to be a strategic mistake.
Rethinking the build vs. buy debate
The "build vs. buy" framing assumes you need to decide on a single approach. This is shortsighted. Instead, businesses should embrace modularity, viewing AI as a spectrum of integrations.
For instance, why invest heavily in building a recruitment algorithm when you can buy a standard model and outsource specific customizations? Similarly, why buy pre-built analytics tools if they restrict how you leverage proprietary data? Businesses should focus on "borrowing" - cloud AIs, open-source solutions, and temporary integrations until there’s a clear competitive advantage in committing to a solution. Basically a temporary setup to test the value before committing. At Algorithma, we use the term borrowing to denote a more non-committal starting point for your AI use-cases - basically where the process of dismantling/returning is easy and cheap.
“AI isn’t a binary choice between build or buy—it’s a spectrum of decisions where modularity and adaptability are key.”
- Simon Lanngren, Management consultant
The real cost of AI
Businesses often underestimate the true cost of data and AI, especially in-house development. It’s not just about upfront investments in talent and infrastructure. Maintenance, retraining models, and integrating ever-changing data pipelines can cost more over time than initially projected.
For off-the-shelf solutions, the hidden cost is flexibility. These tools often work well for initial use cases but scale poorly or fail to adapt to changing business needs. Vendor lock-in can make switching prohibitively expensive, leaving businesses tied to suboptimal solutions.
The cheapest option is to do less with AI, but do it better. Instead of chasing the full spectrum of AI possibilities, focus on a single use case that will deliver meaningful returns.
Misconceptions about data
The idea that “more data is better” often leads to inefficiency and wasted resources. Instead of investing in exhaustive data audits or upfront analysis, start strategically by focusing on what’s immediately actionable. For most use cases, smaller, high-quality datasets aligned with your current goals can outperform expansive collections of irrelevant information.
For example, dive into what massive SharePoints do to implementation of co-pilot in The cost of data: A critical hurdle for co-pilot implementation.
Synthetic data offers a practical starting point, enabling businesses to simulate scenarios and train models quickly while bypassing privacy or regulatory hurdles. By using targeted datasets or generating synthetic alternatives, you can accelerate development without getting bogged down by data complexity.
Success doesn’t require analyzing everything at once. Instead, it depends on refining what you already have and applying it to specific, high-impact use cases. Build momentum by solving one problem well before scaling your data strategy.
Successful data strategies - and cost-effective implementations depend not on quantity, but on thoughtful refinement and purposeful application.
Talent and expertise: Rethinking AI teams
Most businesses approach AI talent as a "build it and they will come" problem, assembling expensive teams of specialists. However, as highlighted in Navigating the age of AI: Rethinking team structure, leadership, and change management, the real challenge lies in integration. Siloed teams, no matter how technically skilled, often fail to deliver impactful results because they lack alignment with business needs and workflows.
The solution lies in fostering hybrid teams that combine domain expertise, AI specialists, and increasingly, digital colleagues. At Algorithma, we talk more and more about the concept of recruiting digital digital colleagues, i.e. AI systems like generative AI are not just tools (or systems/use-cases) but collaborative team members. These digital colleagues augment human capacity by working tirelessly, handling repetitive tasks, and synthesizing complex data, freeing humans to focus on strategy and innovation.
Moreover, outsourcing AI expertise or relying on external resources is not a weakness—it’s a pragmatic strength. By maintaining flexible access to cutting-edge capabilities, businesses can stay current with evolving technology while ensuring their teams remain adaptable.
Building impactful AI teams requires more than technical talent. It demands a rethinking of roles, an openness to collaboration with AI, and a leadership focus on integration and adaptability. Businesses that embrace these hybrid models will not only enhance team productivity but also unlock the full potential of their AI investments.
Why ROI metrics alone may AI be misleading
ROI is often hailed as the ultimate measure of AI success, but it barely scratches the surface of AI’s potential. Traditional ROI calculations fail to capture the strategic advantages AI brings- like enabling faster pivots, reducing reliance on outdated systems, and unlocking entirely new opportunities. These aren’t just numbers on a spreadsheet; they’re the foundation for staying competitive in a dynamic market.
Even more critical is the cost of inaction. Companies that get stuck in endless business case discussions risk falling behind competitors who are already building momentum. The reality is that initiating AI projects - even at a small scale - creates invaluable learning, helping businesses adapt faster and make smarter decisions as they scale. At Algorithma, we call this the ‘game changing mindset’, where one small starting idea sets off a sequence of innovation. Waiting for a “perfect” case means missing out on the compounding advantage of experience.
AI isn’t just a tool for efficiency- it’s a catalyst for innovation, agility, and long-term relevance. If your AI strategy isn’t driving these outcomes, it’s not just underperforming - it’s actively wasting resources.
Don’t let ROI alone dictate your approach. Think of AI as an investment in resilience and adaptability. Getting started early isn’t just smart - it’s essential for staying ahead.
When AI Isn’t the Answer
Not every business or use case needs AI, and treating it as a universal solution can backfire. Some processes are simply too straightforward, or the cost of collecting and maintaining the necessary data far outweighs the potential benefits. In the midst of AI hype, many companies feel pressured to adopt AI - not because it solves a real problem, but because it’s seen as “modern” or “expected.” This tendency to “AI wash” processes can lead to wasted investments and disappointing results.
Sometimes, the better approach is to improve what already works. Small process optimizations, better analytics, or even modernized traditional tools often deliver faster, more measurable returns than rushing into AI development. Thinking through where AI truly makes sense should be a critical part of business planning. Aligning AI initiatives with specific business challenges ensures that efforts are targeted, impactful, and resource-efficient.
When the need for AI becomes undeniable—whether through scaling complexity or evolving business demands—your business will be in a stronger position to implement it effectively. AI should never be a checkbox exercise. It’s a strategic tool for the right challenges, not a one-size-fits-all solution. Knowing when to say “not yet” is just as important as knowing when to dive in.
AI Strategy: Breaking out of conventional thinking
A forward-looking AI strategy isn’t about following trends—it’s about questioning them and challenging the hype. Here’s a refined roadmap for businesses to navigate AI implementation effectively:
Do less, better: Prioritize one or two high-impact AI use cases that address specific business challenges. Avoid diluting your efforts by spreading resources too thin across multiple initiatives.
Delay commitment, accelerate learning: Resist overinvesting early. Start with open-source tools, temporary integrations, or small-scale experiments to build internal understanding and adaptability before locking into a larger strategy. Early learning cycles are invaluable for setting the foundation for scalable AI use.
Outsource with precision: External expertise isn’t a compromise; it’s a strategic advantage. Partner with specialists to bridge capability gaps and stay current with evolving technology while maintaining flexibility.
Measure agility, not ROI alone: Success isn’t just about cost savings or immediate returns. Evaluate AI’s impact on your business’s ability to adapt, pivot, and innovate in response to market shifts. This agility is the real competitive differentiator.
Be realistic about data: Rethink the “more data is better” mindset. Use synthetic or high-quality streamlined datasets to accelerate development and reduce overhead while maintaining focus on actionable insights.
AI isn’t a one-size-fits-all solution or a checkbox for modernization. It’s a strategic enabler, and the businesses that win are those that approach it with clarity, focus, and a willingness to challenge conventional wisdom.
What this means for the build or buy decision
The traditional "build or buy" debate oversimplifies the complexity of integrating AI into a business strategy. Instead, the question should be: How can businesses combine building, buying, and borrowing to achieve the greatest value with the least risk?
At Algorithma, we use a Borrow-Buy-Build (B/B/B) framework to address specific business challenges with AI. Borrowing leverages cloud AI services or open-source solutions for rapid prototyping with minimal commitment. Buying focuses on pre-trained models and enterprise tools for targeted results and quick integration into workflows. Building involves creating custom solutions when differentiation is critical, ensuring a long-term strategic edge. This spectrum allows businesses to test, scale, and refine AI solutions without overcommitting.
The build or buy decision isn’t binary - it’s a spectrum. Businesses must align their AI strategies with their unique goals, challenges, and capabilities. By combining elements of building, buying, and borrowing, companies can stay agile, reduce risks, and unlock the full potential of AI.
The B/B/B-framework isn’t just a methodology - it’s a roadmap. By starting small, aligning solutions with business goals, and scaling strategically, businesses can use AI as a powerful driver of efficiency, differentiation, and innovation.
The businesses that thrive will be those that treat AI as a journey - where thoughtful planning and selective investments lead to impactful, sustainable outcomes.
