A surprising number of enterprises are investing heavily in artificial intelligence without a clear answer to a fundamental question: what do we actually want AI to do for us? According to insights from a recent roundtable discussion hosted by Deloitte and the Association of Business Service Leaders (ABSL), this strategic vacuum represents one of the most significant obstacles to successful AI adoption, even as organizations rush to deploy the technology .

Why Are Companies Building AI Without a Strategy?

The problem isn't a lack of interest or investment. Rather, organizations often find themselves caught between two competing impulses: the desire to adopt general-purpose AI tools like large language models (LLMs), which are AI systems trained on vast amounts of text data to understand and generate human language, and the need to solve specific business problems with tailored solutions. This tension leaves many companies oscillating between vague ambitions and fragmented pilot projects that fail to deliver measurable returns .

"Looking back, many organizations failed to define a clear AI strategy, either seeking mass usage without purpose or attempting to solve everything with a single tool. The fundamental question of 'what do we want to achieve with AI?' was often overlooked," said Eszter Lukács, GBS Advisory and Finance Transformation Director at Deloitte Hungary.

Eszter Lukács, GBS Advisory and Finance Transformation Director at Deloitte Hungary

This strategic gap has real consequences. Without clarity on objectives, companies struggle to allocate resources effectively, measure success, or justify continued investment. The result is what experts call "fragmented efforts and unrealized potential," where excitement about AI technology fails to translate into tangible business value .

What Does a Real AI Strategy Actually Look Like?

Organizations that have moved beyond experimentation share a common pattern: they start with governance and ownership before deploying technology. Many are establishing dedicated AI councils comprising IT, compliance, and automation experts tasked with laying groundwork for a cohesive strategy and responsible AI usage. These councils address complex foundational questions like data ownership, ethical deployment, and alignment with business objectives .

The most successful implementations follow either top-down mandates from executive leadership, common in financial institutions and technology companies, or organic, bottom-up initiatives that empower employees to experiment and innovate. Some organizations are using internal "AI accelerators" and hackathons to foster adoption, demystify the technology, and generate ideas from the ground up .

How to Build an AI Strategy That Actually Delivers Results

• Define Clear Objectives First: Before selecting tools or platforms, articulate what business problems you want AI to solve. Are you optimizing pricing, detecting fraud, automating document processing, or improving inventory management? Specificity matters.
• Establish Governance and Ownership: Create an AI council or steering committee with representatives from IT, compliance, and business units. This group should define policies, data ownership rules, and ethical guidelines before implementation begins.
• Distinguish Between Tool Types: Understand that large language models excel at language understanding but can "hallucinate" or generate plausible-sounding but false information. For high-accuracy applications like financial forecasting or contract analysis, smaller, specialized models or traditional machine learning may be more appropriate.
• Plan for Data Quality and Integration: Ensure your data strategy aligns with AI initiatives. While modern AI can process unstructured data, standardized, high-quality data significantly enhances accuracy and reliability across enterprise systems.
• Invest in Talent and Change Management: Technical skills like Python programming are important, but so are soft skills like problem-solving, resilience, and adaptability. Organizations must invest in both upskilling existing staff and fostering a culture that embraces experimentation.

Real-world success stories demonstrate what's possible when strategy precedes deployment. Organizations have achieved optimization of pricing and promotions, built systems for detecting fraudulent product returns, automated processing of incoming emails and documents, created AI-driven simulations for process optimization, and implemented intelligent inventory management systems. These projects typically deliver returns on investment within 18 to 24 months .

However, a critical pitfall emerges when companies frame AI initiatives solely around cost savings. This narrow focus can cause them to overlook opportunities for revenue generation, capability enhancement, or addressing existing operational gaps. The most successful organizations view AI as a strategic enabler for growth, not just a cost-reduction tool .

"While AI offers immense capabilities, we must understand its probabilistic nature and the need for human expertise in validation and content curation. Misaligned expectations and inadequate user understanding can lead to significant setbacks, underscoring the vital role of skilled subject matter experts," added Zoltán Páll, Technology and Transformation AI Manager at Deloitte Hungary.

Zoltán Páll, Technology and Transformation AI Manager at Deloitte Hungary

The human factor cannot be overlooked. Misplaced trust in AI systems, or conversely, complete refusal to engage with the technology, can undermine even well-conceived projects. The most effective approach positions AI as a tool that augments human capabilities rather than replacing them entirely. This requires robust involvement from subject matter experts in validating AI outputs and continuously refining systems based on real-world performance .

An often-underutilized enabler is access to research and development incentives and governmental support. In some regions, AI-related development involving experimentation or novel combinations of existing technologies can qualify for R&D tax incentives and cash grants, potentially allowing organizations to recover 20 to 50 percent of eligible costs. These mechanisms can materially improve business cases and enable projects that might otherwise struggle to secure internal funding .

The consensus among global business services leaders is clear: the journey toward optimal AI integration is dynamic and requires continuous learning and adaptability. The speed of AI development necessitates infrastructure that can evolve, accommodating future advancements rather than solely addressing current needs. Organizations must foster a mindset that embraces change, encourages experimentation, and views AI as a powerful enabler for strategic growth and innovation, well beyond its immediate cost-saving potential .