AI Strategy

AI strategy is about having a clear view of how to enable the organization to utilize AI capabilities. An AI strategy should encompass the following subsets of strategies.

As of 2025, roughly 95% of AI projects fail. This staggering failure rate reveals a fundamental truth: understanding how to create value from AI matters more than having brilliant data scientists or cutting-edge algorithms. The real challenge lies in wielding AI as a strategic tool rather than chasing technological novelty.

As highlighted in Data Strategy, strategy is fundamentally about making choices and trade-offs—deliberately choosing to be different. AI strategy demands the same disciplined thinking. The question isn’t “What can AI do?” but rather “What should AI do for our organization?”

The Three Pillars of AI Strategy

Effective AI strategy rests on three interconnected foundations. Think of them as a three-legged stool—remove any one, and the entire structure collapses.

The Data Strategy forms the foundation. Without reliable, accessible, and well-governed data, even the most sophisticated AI models become expensive experiments. This means establishing data governance frameworks, ensuring data quality, and creating systems that make information discoverable and usable across the organization.

The Machine Learning Strategy handles the technical implementation. This covers everything from choosing the right infrastructure and platforms to establishing MLOps practices that can reliably deploy and monitor models in production. It’s the engineering discipline that transforms experimental models into business-critical systems.

The Data Science and Analytics Strategy bridges business and technology. This is where methodologies like the ML Design Sprint become invaluable—providing systematic approaches to identify problems worth solving and metrics worth optimizing.

Building an AI Strategy Canvas

Every organization needs a framework for thinking about AI strategically. The canvas should address three fundamental questions.

First, what’s your vision and objectives? This means understanding how AI enables your core business goals, what competitive advantage it provides, and most importantly, what specific metrics will measure success. Vague aspirations like “become more data-driven” don’t constitute strategy.

Second, what capabilities and resources do you have? Honest assessment of data maturity, technical infrastructure, human capital, and available resources determines what’s actually possible. Most organizations overestimate their readiness and underestimate the foundational work required.

Third, how will you implement? Following frameworks like PRIDE (Purpose, Relevance, Inspiring, Deliverable, Enabling) from strategic planning helps ensure initiatives remain grounded in business reality while maintaining organizational momentum.

From Strategy to Execution

It’s a good place to start with an actual strategy. It’s important to understand that strategy is something a little bit different from whishful thinking. Strategy involves actually understanding limitations. You have a company, and it’s about understanding the weaknesses you have as a company. And ALL companies have weaknesses.

That’s often where many traditional businesses, such as non-tech or heavy industries, fall short. They have processes that are not really optimized for AI and machine learning, and they don’t have the knowledge or technical competencies to adapt. Often, they are not led or influenced by highly skilled technologists.

From the journey of traditional IT businesses, we saw that when traditional business types—people with MBAs or degrees in logistics and other forms of leadership—did not understand modern IT technology. So they struggled to adapt. Startups led by technologists, once the got rolling, managed to capture the imense value of IT and the internett. Over time, a lot of traditional businesses absorbed the learnings from these technology-led companies and managed to capitalize on IT through osmosis. (Or, that was the point anyways.)

The core issue is that if you don’t have technology ingrained in your DNA as a manager, leader, or CEO, it’s very difficult to develop a good strategy. Maybe that’s where consultants come in—they can help implement good strategies because they’ve worked with many different technology companies.

However, the downside is that consultants may have helped the same traditional businesses who didn’t understand new technology, simply transferring the same old, broken lessons without the internal knowledge that comes from working with something every single day.

The key insight is that AI strategy succeeds through disciplined execution of fundamentals, not breakthrough innovations. As noted in Data Strategy, ideas are worth nothing unless executed—execution is worth millions.

Links

Strategy Foundations

• Data Strategy - Comprehensive framework for data strategy implementation
• Good Strategy Bad Strategy - Essential principles for effective strategy development
• Driving Digital Strategy - Digital transformation and AI strategy alignment

AI & Machine Learning Resources

• AI Overview - Comprehensive guide to artificial intelligence and machine learning
• Data Science - Practical data science implementation and best practices
• ML Design Sprint - Systematic approach to ML project initiation
• Model Evaluation - Frameworks for assessing AI system performance
• Large Language Models - LLM strategy and implementation considerations

Implementation & Management

• Management - Leadership principles for AI initiatives
• Agile - Agile methodologies for AI project delivery

Thoughts

• It is more important for business stakeholders to focus on viable business use cases and make measurable KPIs than finding out what machine learning can be utilised for. Machine Learning is just a means to an end, not the end goal itself. Let the machine learning engineers figure out the way there, that is what they are paid to do.
• Machine learning projects fail because there are no actionable effects from the output. The output of an AI project in itself is worthless; it is the business value that matters.