Tailor AI Strategy to Your Organisation's Needs

Many organisations are keen to harness artificial intelligence, yet a significant number struggle to move beyond initial pilot projects. The challenge often isn't the AI's capability itself, but a fundamental misalignment between a company's AI ambitions and its operational realities. This includes everything from supply chains and manufacturing processes to existing technology infrastructure and organisational culture.

Consider two contrasting examples from 2018. General Motors (GM) used generative design software to create a lighter, stronger seat bracket. While technically impressive, the design, with its complex, lattice-like form, couldn't be manufactured by GM's traditional stamped-steel supply chain. Retooling would have taken years, so the innovation stalled. Conversely, Apple began experimenting with AI-optimised metalenses for cameras, integrating machine learning with materials science and semiconductor manufacturing. Within two years, Apple had filed numerous patents and was reportedly preparing to embed this breakthrough into its Face ID sensors. The key difference? Apple possessed the integrated system necessary to execute its bold idea.

These stories highlight a crucial point: success with AI isn't solely about the technology. It's about how well that technology integrates with an organisation's existing framework. Industry surveys confirm this, with a significant percentage of companies citing poor cross-functional fit and the need to adjust workflows as major barriers to AI adoption. Many AI initiatives fail to deliver tangible business value, with a substantial portion of proof-of-concepts never making it to production stages.

Understanding the AI Innovation Landscape

To help companies navigate this, we can examine two critical dimensions that influence AI success: value-chain control and technological breadth. These dimensions help define a company's strategic position and inform the most effective approach to AI innovation.

Value-chain control refers to an organisation's influence over its product or service journey, from conception to market. Companies with high control, like Samsung, can rapidly test and scale AI innovations because they manage everything from chip fabrication to retail. Those with low control, such as tier-two suppliers, are more reliant on external partners, making rapid innovation more challenging.

Technological breadth describes the range and interconnectedness of technologies a company must integrate to compete. High-breadth sectors, like autonomous vehicles or life sciences, require AI to coalesce with sensors, robotics, cloud computing, and more. Low-breadth industries, such as food processing, often use AI to refine existing, more stable technology stacks, focusing on process optimisation rather than fundamental redefinition.

These dimensions aren't static. A company might have high technological breadth in research and development but low breadth in customer engagement. By understanding where they possess the strongest leverage, businesses can apply AI with greater focus and confidence.

Four Strategic Approaches to AI Innovation

Our framework identifies four distinct strategies for AI innovation, each suited to different organisational realities.

1. Focused Differentiation: Sharpening Your Niche

Companies with limited value-chain control and low technological breadth thrive on focused differentiation. They operate in mature industries but possess deep expertise in a specific part of the value chain. Here, AI isn't about redesigning the entire system, but about optimising products or processes within a defined domain. It's about going deep, not broad, to achieve precise, high-impact improvements.

Take PepsiCo, which used AI in its potato supply chain to help farmers optimise irrigation and fertiliser use, leading to higher yields and reduced carbon footprints. Similarly, McCormick & Company partnered with IBM to create SAGE, an AI system trained on decades of sensory data to accelerate flavour development and boost new product sales. Fonterra, the New Zealand dairy giant, precisely applied AI to predict milk quality at the farm gate, optimising collection routes and reducing waste. These examples show how AI can fine-tune existing operations for significant gains.

The primary risk here is over-ambition. Zillow's home-flipping initiative, Zillow Offers, famously failed when its AI-derived "Zestimate" pricing model proved inaccurate for off-market listings, leading to substantial losses and the cancellation of the entire business segment. This highlights the danger of expanding beyond one's core competencies without adequate systemic support.

2. Vertical Integration: Wiring the Machine

Organisations with strong value-chain control but relatively limited technological breadth are ideal candidates for vertical integration. By embedding AI into their owned processes, they can achieve substantial impact. AI can link insights across internal systems, revealing efficiencies between data, departments, and processes. This approach is about operational excellence, enabling predictive maintenance, dynamic pricing, and demand-driven logistics. Scale acts as a multiplier, where even small efficiency gains yield significant cumulative benefits.

JD.com, the Chinese e-commerce giant, integrated AI across its logistics network, optimising everything from warehousing to delivery routing. During the pandemic, this intelligent system rerouted deliveries based on containment zones and dynamically reassigned inventory, allowing JD.com to maintain uninterrupted service. Likewise, ExxonMobil used AI to interpret seismic data, cutting well-drilling times by 15% and saving millions. Walmart, with its extensive control over its supply chain and operations, employed AI to anticipate demand spikes, rerouting emergency supplies ahead of Hurricane Ian.

These examples underscore how connecting the dots across a tightly controlled system using AI can create a unique competitive advantage. However, even strong operations can stumble if ambition outpaces execution. GE's Predix platform aimed to be the "Microsoft of industrial AI" but ultimately scaled back due to siloed data, internal resistance, and a lack of coherent software integration.

3. Collaborative Ecosystem: Working the Network

For companies operating in technologically complex ecosystems but lacking full control over market reach, forming a collaborative ecosystem is key. Success stems from strategic partnerships, sharing innovation risk, infrastructure, and expertise. These companies often work in fast-moving, high-tech sectors where execution depends on regulators, researchers, or platform partners. Their advantage lies in shared platforms, co-developed tools, and alliances that align incentives.

Novartis and Microsoft, for example, formed an AI innovation lab to accelerate drug discovery, using machine learning to predict molecular behaviour and optimise clinical trials. BMW Group's alliance with Intel and Mobileye developed autonomous driving solutions, with each partner contributing distinct capabilities. Perhaps most strikingly, the partnership between Pfizer and BioNTech during the Covid-19 pandemic saw BioNTech's AI models screen thousands of mRNA candidates, while Pfizer's manufacturing and regulatory capabilities accelerated production. This quadrant is adept at unlocking fundamental discoveries that can reshape science or technology.

The challenge here lies in coordination. IBM's collaboration with MD Anderson to revolutionise cancer care with Watson-powered AI struggled with organisational and integration challenges, failing to move beyond the pilot phase. Effective partnerships require deep alignment of knowledge, incentives, and execution, not just technological synergy.

4. Platform Leadership: Shaping the Norms

At the highest levels of both technological breadth and value-chain control are platform leaders. These companies don't just adapt to change, they define it. They build infrastructure and ecosystems, setting standards and creating platforms that others want to build upon. This is the domain of orchestration.

Bloomberg's launch of BloombergGPT, a finance-specific large language model, exemplifies this. Trained on vast financial datasets, it summarises earnings reports and assists with risk modelling within Bloomberg's integrated terminal, establishing a new industry standard. Siemens Healthineers holds a similar position in medical imaging with its AI-Rad Companion suite, which integrates directly with hospital systems to analyse scans and improve diagnostic accuracy. Microsoft, with its Azure OpenAI Service, GitHub Copilot, and embedded AI in Microsoft 365, provides the enterprise backbone for generative AI, building leadership through both technical performance and a focus on governance and transparency. Chinese AI models, like those from Alibaba, are also demonstrating significant platform leadership in their domestic markets, integrating shopping features into core AI apps and influencing global derivatives markets. You can read more about China's growing influence in Chinese AI models dominate global derivatives.

These leaders possess the scale, data, and architectural reach to spot and act on weak signals before others. However, this reach comes with increased responsibility. Google's DeepMind Health, for instance, faced public backlash and ultimately lost momentum when it accessed millions of NHS records without proper consent. The failure wasn't the algorithm, but a breach of trust.

The Human Element: Engaging Your People

Beyond the strategic frameworks, the most defining challenge in AI adoption is often human, not technical. Many initiatives falter because employees resist new tools, fearing job displacement. Surveys indicate a significant number of employees actively push back against AI initiatives, sometimes even sabotaging efforts.

Transparency, dialogue, and personal experience are crucial. When Rent a Mac introduced an AI-driven inventory system, initial anxiety led to delays. However, by appointing AI champions to demonstrate real use cases, engagement levels tripled. Similarly, Colgate-Palmolive’s internal AI Hub empowered employees to develop their own AI assistants, fostering buy-in and improving workflows. This highlights a shift in managerial roles: from coordinating people to helping teams collaborate with algorithms, interpreting machine insights, and translating technical progress into human advancement. This requires a cultural shift towards experimentation and real-time learning. More on the changing nature of work can be found in AI creates a new "meaning" of work, not just the outputs.

Ultimately, successful organisations view AI not just as a solution, but as an ongoing question: "How can we work smarter, together?" Companies like P&G demonstrate this holistic approach, operating across all four quadrants. They apply AI with precision for immediate value, integrate it where scale drives performance, partner where complementary capabilities are needed, and create platforms to shape key ecosystems.

The next decade of AI success will hinge not on the number of pilots launched, but on the ability to scale. This means aligning ambition with organisational reality, empowering people, and ensuring AI serves the broader strategy. AI is a tool to bring strategy to life. The difference between GM and Apple wasn't just the technology; it was the entire system surrounding it. Companies that build the right systems, where ambition meets execution, will be the ones to lead. For further insights into delegating tasks effectively to AI, consider reading Your AI Agent: 3 Steps to Effective Delegation.

What's your organisation's biggest hurdle in scaling AI initiatives? Share your experiences and insights in the comments below!