Get your data ready for AI

In a rapidly changing world, AI is no longer a desirable beacon of innovation but rather an undeniable necessity, yet not all companies have embraced its potential. There are a ton of reasons for that, but let's focus on one of the biggest hurdles: data.  

The AI Hype vs. Data Reality 

AI promises unprecedented efficiency, predictive insights, and transformative capabilities. It sounds exciting, and even better, it may well be true.   

However, let's not forget that these promises depend on one element: DATA.  

Without quality, well-structured data, the implementation of AI initiatives just makes the mess faster. There will be no magic, just "trash in, trash out".  

A recent SAP blog makes this point well: while AI technologies are advancing rapidly, the success of their implementation depends largely on the quality of the data they consume.  

The AWS 2025 research report Scaling Generative AI for Value revealed that while 83% of chief digital officers acknowledge that gen AI is a top strategic priority, at the same time 52% of the same group rate their data foundation's readiness for gen AI implementation a five or lower on a zero to 10 scale (where zero is "not at all ready," and 10 is "completely ready").  

Why is it so? I would like to share some personal notes on the data dilemma that businesses may face.  

Despite the importance of data quality, many organisations face challenges: 

1. Outdated infrastructure: Outdated systems may not support the data requirements of modern artificial intelligence applications. 
2. Source of truth: Companies often lack a single, authoritative repository or system where an organisation stores and maintains its most accurate and reliable data. 
3. Lack of data governance: Without standardised policies, data quality and consistency suffer. 
4. Data silos: departments often work in isolation, leading to fragmented data storage and management. 

Another big question many companies have is: What should we do with historical data? How much historical data do we need to "train" AI? The short answer is: It depends on the situation, but more is not always better. 

The universal rule of thumb is that relevance is more important than volume. Just because you have 10 years of data doesn't mean it's useful. Outdated, unstructured records are a disadvantage rather than an asset.   

That said, well-stored historical data can give AI systems richer context, especially in industries like finance, manufacturing or logistics where seasonality and long-term trends are important. 

If you're using predictive AI (for example, to forecast sales or identify risks), historical data is more important. If you're using generative AI (e.g., chatbots or content generation), real-time data and context typically play a bigger role. However, it's worth noting that GenAI modules can also be built on top of predictive models, meaning they can effectively leverage historical data as well.   

Moderna's case: AI and Data  

Let's look at Moderna, a Cambridge, MA-based biotech company that made a name for itself in 2020 upon releasing one of the first COVID-19 vaccines approved for emergency use by the U.S. Food and Drug Administration.  

It has emerged as a leader in AI-driven biotechnology, integrating artificial intelligence into all drug development processes. 

Working with OpenAI, the company has developed more than 750 specialised GPT tools, such as "Dose ID," to speed decision-making in vaccine development. 

Moderna has also launched an internal Artificial Intelligence Academy to enhance employee skills and foster a culture of innovation within the company. This digital shift allows the company to scale significantly without the need for a massive increase in staff. According to CEO Stephan Bansel, with artificial intelligence, "If we had to do it the old biopharmaceutical ways, we might need a hundred thousand people today. We really believe we can maximise our impact on patients with a few thousand people, using technology and AI to scale the company."  

So, what is the strategy for Data Readiness? 

To overcome the gap between the desired benefits of AI and reality, organisations should consider the following: 

• Data Audit: the organisation should understand its current state: where data is stored, its quality, relevance and consistency. 
• Data governance: Set in place clear policies and procedures for data governance and security. 
• Data infrastructure: Think about investing in the platforms and tools that support the real-time processing, scalability and integration required for AI applications. 
• Data-centric culture: Oh, it's the hard one. Building a data-driven culture is a gradual process, not a one-time effort. You can find more about nurturing and developing a data-driven culture in the not-new but value-packed article "The Four Key Pillars To Fostering A Data-Driven Culture" By Brent Dykes.  

AI at SAP 

SAP has been building AI into its platforms for years. Whether you're running SAP S/4HANA, using SuccessFactors, Ariba, or the Business Technology Platform (BTP), AI is becoming an integral part of the system. 

Here are just a few examples: 

• SAP Joule is a digital assistant embedded in SAP systems. It optimises tasks, extracts information and recommends actions by understanding the business context. 
• AI in SAP S/4HANA - automating invoice processing, predicting payment delays or optimising supply chain scenarios. 
• SAP BTP AI Core - for those who want to build their own AI, integrate open source models, or deploy finely tuned solutions at scale. 

Conclusion  

The future of AI is huge, but let's be real: it all comes down to your data. Think of it this way–if AI is the king, your data is the entire kingdom. To actually get AI working for you and delivering real results, you need solid data and a smart strategy. Ready to explore how your data can truly power your AI initiatives? Contact our experts via eleks.com 

Written by Gaiana Karakashian (MBA) , SAP Manager at ELEKS.