Many organisations see AI as a potential route to greater productivity. But when I look across the client base here in the South, I see a varied picture, which may reflect the wider situation nationally.
Three AI phases…
Organisations generally fit into one of three buckets, along a 60-35-5 split. In the 60% camp are those businesses who are using AI tools like Copilot that are part of their software suite, but adoption is mixed. Some team members are using it extensively and learning fast; others are more resistant and may hardly be using it at all. In other examples, businesses are still working through their governance and adoption permissions, perhaps due to regulatory aspects and this slows adoption. The challenge is that because this is so new, there is no clear North Star. It is clear from my conversations over the last year or so that there is already a lot of latency in terms of capabilities that are not being leveraged from software that businesses have already invested in. AI has become another example. Businesses are overall making progress, but it is relatively slow and uneven.
Then there is the 35%. Here, businesses are taking more active ownership of AI. They have thought through the risk and governance issues and developed an approach they are comfortable with. They have started building their own agents, they are focusing on prompt engineering, and they have developed use cases – often across three core areas: finance, HR and sales. In some sectors, such as technology, financial services and consumer and retail where technology capabilities are already advanced and the business is data-rich, there are likely to be active use cases across all three of these areas. In other sectors, it may only be in one or two based upon the specific use case and business benefit. These enterprises are developing their use of AI quickly.
Finally, we have the 5%. These organisations are in the vanguard. They are likely to be using advanced AI models such as (Anthropic’s) Claude and (Google’s) Gemini, building their own bespoke tools off the back of them. They are not only using AI for back-office efficiencies but for other benefits such as scenario analysis, simulation and modelling (digital twins) and deployment in front office activities.
Critical differentiators
But what influences how much progress a business is making in its AI journey? I see four key differentiators:
Some CEOs or founders have made it an absolute imperative that the business embraces AI. They may have held off-site days with senior management to drill down into how AI can be implemented in the business. This creates a clear mandate to go back to the office, adopt it, use it, experiment with it and cascade it across teams. This leadership drive is key to creating and building momentum.
At one point, there seemed almost to be a belief that data management, e.g. data lakes, didn’t matter because AI can handle unstructured data. This expectation has not matched the reality. In fact, a strong master data management strategy is essential – companies who are on top of their data are in a different zone from the rest. Taking the time to ensure data systems are robust is fundamental.
Don’t expect AI outputs to be perfect – they still need human validation and oversight. That due diligence is key. An AI model can save you huge amounts of time and work if the prompts you give it are sound, and hallucinations have reduced as models have developed. AI can take you a significant amount of the way. But that last amount still comes down to the skill, knowledge and experience of your people and their areas of expertise. It’s only the same as using junior, less experienced staff to support on activities and that you consider proportionality. For example, you may not be as worried of hallucinations on lower level tasks but you wouldn’t use AI as the only tool to assess bidding for a large contract or the offer price for an M&A target, where getting it wrong can cost millions.
Businesses run on lean operational models – everyone is multi-tasking; everyone is short of time. So, how and when are your team members expected to get up to speed with AI? Just as much as it is a technology issue, adopting AI is also an L&D issue – that’s the realisation I’m hearing from business leaders around the South. Chief People Officers, Heads of Learning & Development or equivalent have a crucial role to play in devising the strategy for training, upskilling and self-led learning on AI. Along with this, you need to instil a culture where colleagues share tips and insights on AI techniques. That needs to become a common conversation. The danger otherwise is that we’re all using AI as a personal assistant but we’re not seeing outside our own usage: it’s a vertical, one-to-one conversation that we’re all having, when opening it out across team members will give people a much richer frame of reference.
Progress in the South
Overall, I am encouraged by the progress that I’m seeing businesses making. In the South, the Thames Valley in particular plays host to a myriad of tech-enabled businesses who are embedding AI into their products & services and the ways in which they get work done.
Getting tangible results and ROI from AI is not straightforward – it’s a whole new science after all – but driven by the boardroom, I expect more businesses to move up the adoption curve and make AI part of the new normal.
Our AI and machine learning insights
Something went wrong
Oops!! Something went wrong, please try again
Get in touch
Discover why organisations across the UK trust KPMG to make the difference and how we can help you to do the same.