Getting data right in the age of AI

Georgia Tangalakie at SAP discusses the unintended consequences of AI-driven shadow IT and why businesses need a structured approach to data management

 

Concerns about the pace of technological advancement are no longer theoretical. As powerful technologies such as Artificial Intelligence (AI) become more accessible, the focus is shifting from what AI can do to whether it is being used with the right level of oversight and responsibility. What were once abstract debates are now urgent questions facing every organisation.

 

The widespread accessibility of AI tools has accelerated the rise of shadow IT, with employees bypassing traditional governance structures to deploy powerful, self-service AI applications. This creates an environment where businesses must grapple with maintaining control as unmanaged AI systems begin making critical decisions based on fragmented, unverified data sources.

 

Some organisations are increasingly deploying transformative technology without fully understanding the risks or putting proper safeguards in place. The pattern is clear: impressive capabilities are being rolled out without adequate oversight, governance or clarity around consequences.

 

The question is no longer whether AI will transform business operations, but whether organisations and employees can harness its power responsibly.

 

 

AI-powered shadow IT hidden in plain sight

Shadow IT is not a new challenge, but AI takes it to a new level. With so many generative AI tools now readily available, employees can solve problems, generate content, or make recommendations at speed. This happens often without needing any technical expertise or approval.

 

This speed is both a blessing and a risk. In their enthusiasm to experiment and move fast, teams often pull data from disparate sources, bypassing enterprise-grade controls in favour of quick, isolated fixes. Over time, these short-term solutions accumulate, and organisations are left with a patchwork of systems, models and insights that don’t speak the same language.

 

The risk isn’t just that teams are duplicating efforts or misinterpreting data. Business-critical decisions affecting customers, supply chains, product development and strategic direction are increasingly being made based on unverified siloed information. When AI systems operating on flawed data foundations make recommendations that influence growth strategies, the potential for bias or error multiplies exponentially.

 

 

Establish confidence through unified data

The antidote to this growing risk isn’t to clamp down on experimentation. It’s to build the right data foundation, one that supports innovation while maintaining context and integrity.

 

This means giving employees access to high-quality, AI-ready data from across the business. It’s essential to build one harmonised layer that connects all business AI applications and ensures that everyone from developers to decision-makers can rely on a single source of truth. This foundation keeps context intact, so the entire business can see where, how, when and why data was produced, building trust and accurately informing decisions. When data is unified, it also supports regulatory demands and keeps the business agile to future compliance requirements.  

 

 

How data silos lead to inefficiency

There’s a significant cost benefit to this too. When growth is the unanimous business goal, organisations cannot afford to haemorrhage spend on an inefficient IT landscape.

 

It’s estimated that organisations today spend up to 50% of their IT budgets on data and analytics, with a significant portion of that going to attempts to harmonise disconnected data sources. Yet, despite these efforts, many businesses still lack a continuous, unified data layer that brings these sources together in a coherent, usable way.

 

That’s not just inefficient, it’s a missed opportunity. In the age of AI, the power of data lies not just in how much you have, but in how well it’s connected. Without a shared foundation, AI models risk drawing the wrong conclusions or being trained on outdated information. This in turn leads to additional budgetary pressures. Businesses need to confidently scale AI across functions, knowing insights are accurate, secure and compliant.

 

 

Connecting data to real business values

To move from raw data to real business outcomes, organisations need more than just infrastructure. They need a strategic approach to data and analytics that supports decision-making at every level.

 

This means combining new technologies with existing business processes to create enriched, curated data products that deliver meaningful value. It means equipping users with advanced analytics, benchmarking tools and AI-powered insights applications that can both interpret the data and recommend actions.

 

This strategic approach also helps limit the spread of shadow IT by reducing the need for employees to seek out unapproved tools or shortcuts. By aligning data initiatives with established governance frameworks and cultural values, organisations can ensure consistency, compliance and trust in the data being used. At the same time, it creates space for innovation and agility, enabling teams to move quickly and confidently within a well-defined structure.

 

When done right, the benefits are clear: smarter decisions, faster responses and better outcomes across the board.

 

 

Establishing AI confidence

The real question isn’t whether businesses can use AI, but whether they’re prepared to use it responsibly. True readiness goes beyond experimentation. It demands strong data foundations, clear governance and teams equipped with the right tools and guidance. It also calls for a culture that supports innovation while maintaining the right safeguards. 

 

Innovation alone is not enough. When ambition outpaces oversight, risks multiply. Lasting transformation depends on structure, control and accountability. That balance is what turns potential into meaningful, sustainable progress.

 

---

 

Georgia Tangalakie, Head of BTP UKI at SAP

 

Main image courtesy of iStockPhoto.com and Vertigo3d