Paul Moxon, VP of Data Architecture & Chief Evangelist at Denodo, argues that data is the real foundation of AI success – and that overlooking it might render AI investment a waste of time.

AI is undoubtedly going to be transformational for organisations everywhere. The sector is predicted to reach $407 billion by 2027 and grow annually by over 37% between now and 2030. In 2023, AI software already accounted for $70 billion in revenue, while McKinsey reports that an increasing number of software and data engineers are being onboarded for support.

The popularity of AI is not in question. Nor is its economic impact. There’s little surprise that it is now at the heart of many conversations about technology happening in all kinds of business.

However, there is a real possibility that many of those conversations are, to put it bluntly, a waste of time.

That’s because any conversation about Gen AI is really a conversation about data, and so the question of data has to be the starting point for any approach to implementing AI, maximising its capabilities, and measuring its success.

Or, as Warren Buffet once said, “price is what you pay; value is what you get” – and too many AI initiatives put the financials ahead of the fundamentals that need to be in place for Gen AI to deliver value.

Where the value of Gen AI lies

The huge promise of Gen AI rests on its ability to deliver in two main roles: improving the quality and understandability of data and reducing the barriers to information for users who don’t have technical skills. However, without an underlying logical data architecture, harnessing its capabilities will be challenging. This is because Gen AI’s greatest challenge is data, and the ability to read and understand that data.

It is critical that information delivered via AI chatbots is not only secure and correct, but also that it is contextualised within the wider business landscape. Providing the most relevant and accurate contextual data to the Large Language Model (LLM) is crucial if organisations are going to realise the full benefits of Gen AI. Making sure that data is AI-ready is a step that organisations can’t afford to miss.

This is an urgent cause. In McKinsey’s Global Annual Survey 2024, 65 percent of respondents claimed that their organisations are regularly using Gen AI, and three-quarters predicted that it will lead to significant change in their industries. While many say they have already experienced an increase in revenues and a decrease in costs as a result, investments at this scale need to be anchored in a sound strategy if businesses are to see the returns they need.

Setting the scene for effective implementation

LLMs that draw on enterprise information to allow employees to just ask questions about the business, such as corporate policies, employee deliverables or inventory, will be invaluable to any organisation. Private corporate data is often far less accessible, however, due to limitations on search functionality. That’s where retrieval augmented generation (RAG) steps in, enhancing LLMs with vetted contextual data to enhance the accuracy and relevance of their generated outputs.

Effectively, RAG combines the LLM’s ability to produce readable and insightful text with the trustworthiness and usefulness that comes from the data that businesses rely on in their existing workflows.

The first step towards that is to unify disparate sources of data from across the organisation for a consolidated view that enables a cohesive and accessible approach to the data stored. Using a data fabric – an abstraction layer that connects data sources through metadata – teams can facilitate Gen AI’s access to timely and accurate information that can be shared through LLMs.

That bridging of data through a data fabric, underlined by data virtualisation and the ability for RAG to provide contextually relevant data, uncovers significant improvements for LLMs, enabling people to access knowledge more effectively and efficiently, while also removing barriers to working with that data for people who don’t have specialist data skills.

Generative AI offers a significant opportunity to innovate and succeed in a competitive market, if businesses implement a strategic approach to data management. Like chess, AI requires a long-term strategy with careful planning, data optimisation and refining algorithms. Over time, the value compounds leading to significant innovation, efficiency and leadership. This means that an investment in getting not just the Gen AI tool itself, but also the wider business, its data, and its operations right will be key to Gen AI success.