As we move out of the hype curve for every new and disruptive technology, people tend to fall into the disillusionment phase. It’s normal, and it has happened before. But this time feels a little different.
The scale of general AI use is exceptional. Even if you take something as mainstream as “search”, we are already witnessing profound changes in general use due to this technology. With breakthroughs like:
- Google Gemini: Integrates AI-generated answers within Google Search, aiming to deliver results 40% faster.
- ChatGPT “Search Mode”: Combines AI capabilities with live web data to reduce inaccuracies.
- Voice-AI Fusion: Devices like Alexa are beginning to cross-check AI responses with traditional search engines to enhance reliability.
General adoption numbers are mind-blowing. According to ChatGPT the numbers on its platform look like this:
- Weekly Active Users: 800 million
- Daily Active Users: 190.6 million
- Monthly Visits: 5.72 billion
- Total App Downloads: 64.27 million
- Daily Queries Processed: Over 1 billion
AI use cases are aplenty, and on paper they all show promise. AI is here and it appears to be staying and headed to redefine humanity. However, in the corporate world there’s a somewhat different picture. While there is a huge potential for AI-driven personalised experiences, the reality is that most organisations will struggle, and are struggling to get off the starting blocks.
According to recent reporting from The Register, only about 5% of enterprise GenAI pilots show meaningful business impact. Most never make it past the pilot phase, and even fewer reach production or deliver measurable improvements to revenue or cost.
McKinsey has referenced a lot of issue with what it calls pilot purgatory, which surfaces for a number of reasons. Firstly, most companies aren’t data-ready, they can’t operationalise it or make good, clean and usable data available in the way they need to. Secondly, they aren’t built in the holistic, open ecosystem models needed to run GenAI natively, so their technologies and business architectures struggle to adopt models easily, or at scale. Lastly, organisations aren’t usually as digitsed as the AI world seemed to think.
All of this means that when applying AI to digital, or putting natural language interfaces over connected experiences and so on, it’s still often all in legacy. To escape this purgatory there’s three big things we need to address.
What are we AI’ing?
The critical question that must be asked first is, is this about using AI to optimise AS-IS or is it a chance to rethink what personalised products and experiences look and feel like? For me, it has to be the latter. In AS-IS there’s too much friction in a lot of processes that’s not needed. Once this friction is removed, intelligently orchestrating experiences could be much deeper, driven by knowing your customer and giving them what they need and want in every interaction.
Current legacy IT systems constrain true adaptation, because even with modern elements of architecture, such as partial cloud strategies, some open API based capability, or elements of digital experiences, the collective picture doesn’t change that much. When people look to launch new business models these foundations show up the same as before, and then everyone is back into time consuming IT change requests.
AI is Colluding with Data in Personalisation.
The second question is, is personalisation a challenge or an opportunity? I think it’s both, and until it’s addressed that way we risk patch personalising some things and making experiences fragmented in the process. We don’t just need more data, better data, or even reliable data. We need to operationalise data and how we interface and control AI when using it.
Agentic systems offer some great potential, but they are and should be much more holistic. Equally they are nothing without the right data model to work within. However, AI is just the tool for intelligence and automation. Great examples of personalised experiences that both protect and sometimes improve trust have existed for a long time in many industries.
When looking for relevant examples of high trust, personalised experiences, the best still show up in the eCommerce world. But we can also look at industries like hotel, catering and hospitality where they can often be at the right moment, incredibly personalised and meaningful.
FS companies sometimes misinterpret this – no I don’t need you to celebrate my birthday with me, but some welcomed advice at the right time would be amazing thanks! And when we think about really great embedded insurance moments we often think about when income protection considerations appear in our bank experiences. Too little and, quite frankly, not enough.
Privacy-First is Really Customer-First.
Trust is hard to earn and easy to erode. All too often we see the wayward use of AI, alongside ever increasing security risks. Remembering it’s your customers data and you are only permitted to use it where you can responsibly add value is essential to better, more intelligently orchestrated experiences. And it relies on the other two things to be true.
In most industries outside of eCommerce these paradigms feel hard to get to. That’s because markets like Banking, Financial Services, and Insurance are highly product verticalised, regulated, legalised and built in deep systems on the idea of transaction volume and efficiency. When scaling these industries in technology for the last thirty years this has made sense. Even digital was possible because it was largely treated as a channel, and evolved experience by experience.
Whilst these digital offerings can claim to be in the right technologies, i.e. cloud hosted etc, they aren’t architected to be built around the customer. Being a customer first organisation has to therefore move from the conceptual to the real.
Being truly architected this way means the customer is baked into the “stack” and into the working model of the company. Everyone is then focused on increasing the knowledge of a customer and their ability to act on this knowledge. It’s inherent in the systems design, and will appear naturally in every interaction.
Equally the tools placed around this kind of enterprise design then typically include customer portals, provide digital capabilities (e.g. multiple payment options) or user interfaces that naturally work across any device or setting. Perhaps most importantly the customer will show up in any channel of interaction, making the call centre, face to face, digital and now AI experiences seamless and integrated.
We need to move from systems of record, storing and operationalising static experiences to systems of intelligence increasing the knowledge of a customer and the ability to responsibly act on that knowledge. To do that in the AI era that’s emerging we need to rebuild not just AI what we do today. The priority is to write positive business cases and set paths towards better propositions and future.
The focus must be on not just making data flow, but making data build trust, by driving relationships through new paradigms yet to be imagined. This isn’t an AI revolution, it’s the emergence of an imagination led economy.
