Before organisations can scale AI effectively, they need the right foundations in place across infrastructure, data and their existing technology estates.

Over the past year, I’ve had the opportunity to attend a wide range of industry events with RedRock, spanning a wide range of sectors and technology disciplines. These have included government transformation and the future of digital services, alongside broader topics such as data, user-centred design, cyber security and organisational resilience.

Across all of these themes, artificial intelligence has been central to the conversation in the various panel discussions and keynote presentations I’ve attended. What has particularly interested me over the past year is just how much the conversation has evolved from experimentation and early adoption to implementation at scale and looking towards readiness for future iterations.

It’s been fascinating to observe not only how quickly AI has developed, but the pace at which organisations have been able to adjust and deploy the likes of generative AI and tools such as Claude and Copilot into operations to create time and cost savings, unlock capability and revolutionise ways of working, in such a short space of time.

While many conversations remain centred around testing and learning, I’ve noticed a distinct shift in focus away from testing a new model or tool to laying the right foundations at an organisational level across data, architecture, infrastructure, skills, security and governance, to ensure for effective adoption of AI that aligns with overall strategy.

The importance of data foundations

Data has also been a recurring theme across many of the events I’ve attended over the past 12 months, and increasingly I’ve found that conversations around AI naturally lead back to the quality, availability and accessibility of organisational data.

There is understandably a lot of focus at the moment on the capabilities of the latest AI models, but the quality of the outputs they produce is heavily influenced by the data and information they have access to. I’ve heard various technology leaders express that AI can only be as good as the quality of the data it has access to, and for organisations with large and complex technology estates, this can present a significant challenge, particularly where information is distributed across multiple systems, departments and platforms.

Many organisations are still dealing with fragmented data, duplicated information, inconsistent definitions and legacy systems that were never designed to share information easily and do so across departments. These challenges may not be new, but the increasing use of AI brings them into sharper focus because organisations are looking to connect more information and automate more processes.

This is why I think the conversation around AI-ready organisations increasingly needs to include the less visible work that happens behind the scenes. Improving data quality, establishing clear ownership, strengthening governance and enabling interoperability may not have the same immediate appeal as launching a new AI capability, but they are fundamental to making that capability useful.

In many ways, some of the most important AI investment an organisation makes may not actually be in AI itself, but in creating the data foundations that allow AI to deliver meaningful and reliable outcomes. This all relies on a wider theme that has continually emerged at the events I’ve attended, where the human side of AI adoption in terms of culture and behaviour is crucial to the success of any initiative.

Working with existing technology estates

Another consistent conversation around digital transformation is the challenge of introducing new technology into existing environments. AI doesn’t arrive in isolation, and in the majority of cases organisations aren’t starting from scratch, as most have technology estates that have developed over many years, with a combination of modern cloud platforms, established enterprise systems and legacy technology still supporting critical services.

This creates an interesting challenge as organisations look to introduce AI. Many tech leaders have shared their thoughts on how there can be a temptation to view AI as another layer that can simply be added to the existing technology environment, but in reality the relationship is more complicated.

AI often needs access to data held across multiple systems, integration with existing applications and processes, and the ability to operate within established security and governance frameworks. Where those foundations aren’t in place, the limitations of the existing architecture can quickly become a constraint on what AI can achieve.

From the conversations I’ve listened to, I think this is where enterprise architecture and wider technology strategy have an increasingly important role to play. Rather than looking at AI as a standalone programme, organisations need to understand how it fits within the wider technology landscape, where existing systems can support new capabilities and where targeted modernisation or integration could unlock greater value.

For organisations with significant legacy estates, this doesn’t necessarily mean replacing everything, it means understanding where change will have the greatest impact and creating an environment in which new capabilities can be introduced without adding further complexity.

Looking Forward

Getting the technological foundations right is an essential first step, but infrastructure, data and architecture alone won’t determine whether AI delivers lasting value. The technology may provide the capability, but organisations also need to consider how that capability changes the way people work, how processes are designed and where responsibility sits. As AI moves from experimentation into everyday operations, the focus therefore needs to broaden beyond the technology itself to consider operating models, skills, culture and the outcomes organisations are trying to achieve.

AI readiness is therefore about much more than having access to the latest tools. The organisations best positioned to make use of increasingly capable AI will be those that take a broader view of the foundations supporting it, from infrastructure and data to architecture, integration and the existing technology estate. These may not always be the most visible or exciting elements of AI adoption, but they are critical to creating an environment where new capabilities can be introduced effectively, securely and sustainably.

However, getting these foundations in place is only part of the challenge. As AI becomes more deeply embedded within organisations, it will increasingly influence how people work, how processes operate and how decisions are made. The next question is therefore not simply whether an organisation is technologically ready for AI, but whether it is organisationally ready for the changes that come with it.