“Foundations are key... but if you're building the foundations without the vision, where do you go next?”
Terry McCann, CEO of Advancing Analytics, in conversation with Andrew Phethean, UK & Ireland Industry Lead, Enterprise Growth at Databricks, on why the Data + AI conversation is shifting from experimentation to outcomes, what he expects to hear at the Databricks World Tours, and why the next phase of AI will require much deeper partnerships between the Data + AI platform, partners and its customers.
There’s been a noticeable change in the conversations we’re having with customers. A year ago, there was still a lot of excitement around experimentation, chatbots and simply finding places where gen AI could make someone a little more productive. Now, organisations are starting to ask the much harder question: what value are we actually getting from all of this?
That's something Andrew and I came back to again and again during the conversation. Andrew described the change pretty neatly as moving from “token max” to “value max”. Organisations have experimented, they've invested and, in many cases, they've built multiple versions of similar ideas in different parts of the business - the question now is how you take the ideas that really work, put them into production and create value from them across the organisation.
“The tide has changed in the sense of: what value are we actually getting from it? And now, how do we almost value max rather than token max?” – Andrew Phethean
For me, that's probably the biggest shift in the market: customers aren't asking us to just build things anymore.
If somebody came to see us at an event a year ago, there was a good chance they were looking for a technology partner. They had an idea, they'd decided they wanted to build something and the conversation was essentially: can you help us implement it?
Increasingly, the conversation today is the need for a partner who’s prepared to step back and say: yes, we can build that, but is that actually the right thing to build? What data do you have? What outcome are you trying to create? And is there a bigger opportunity here that we aren't looking at yet?
“We don't just solve the problem that we’re being asked, but help customers critically assess the suitability of the data that they have and the outcome that they might be able to drive from it.” – Terry McCann
To me, that’s the difference between being a technology implementation partner and becoming a genuinely useful strategic Data & AI partner, because you're bringing the technology together with an understanding of the customer's industry, their data and the business problem they're trying to solve. Because of this, the people involved in those conversations have changed as well; we now often have conversations with CAIOs, CFOs and CMOs because the platform touches a much wider set of business outcomes.
That's where I think things become interesting, because the conversation stops being about buying some technology and starts becoming a conversation about changing something in the business.
We've seen enormous investment in generative AI experimentation, but there's obviously a substantial difference between getting a pilot to work and getting an AI capability working reliably inside a large enterprise, at scale, with the governance, economics and organisational change wrapped around it.
Andrew talked about four barriers Databricks is seeing: context, control, cost and choice. He described the importance of these 4C’s:
But there was another point that Andrew made that I think can easily get lost in all the excitement around technology: “It's having the right technology in the right hands for the right tasks.” Andrew Phethean.
That's really important because, for all the extraordinary progress we're seeing in AI, some of the hardest problems organisations face are very familiar: adoption, change management, skills and getting people to use technology effectively.
We can talk about agents all day long, but if the organisation isn't able to operationalise them, or the data underneath them isn't trusted, governed and accessible, having an incredibly sophisticated model doesn't really solve the problem.
I suspect that's why the conversations at Databricks Data + AI World Tours will be interesting this year, because we're getting to the point where the discussion about what the technology can do is becoming inseparable from the discussion about how enterprises actually use it.
Andrew expects governance to be central to those conversations, including Unity Catalog alongside Databricks' broader agentic technologies and applications, with customer examples providing the practical evidence of how organisations are using the platform end-to-end.
And, personally, that's always the part of these events that I enjoy most.
We're fortunate to have a number of our customers taking to the stage at World Tour London this year, including Sega, and although it’s great to hear about the technology they’ve adopted, the more interesting part is seeing how they’re translating the platform into something that really matters for them.
We're also seeing the Databricks conversation expand into business areas and personas that historically might not have viewed themselves as being particularly close to a data platform.
I mentioned Customer Lake when chatting to Andrew because I think it's a good illustration of that. It starts opening conversations with a Chief Marketing Officer, for example, and not just about marketing to retail customers, but about things such as increasing brand loyalty within a bank or identifying opportunities across an insurance portfolio.
That conversation would never have happened years ago!
“The conversation again is shifting away from Databricks being just this data platform to ultimately being the backbone and the real kind of data ecosystem in a lot of our customers.” – Terry McCann
And if that's the direction of travel, it also changes what customers should expect from their strategic partnership.
This might sound counterintuitive, but I think the easier and more capable technology platforms become, the more specialist the technology partner needs to be.
As the barrier to adopting technology gets lower, customers don't necessarily need armies of engineers to implement the solution. Instead, they need help in those difficult areas where technical expertise has to meet domain knowledge.
That's where I see the role of partners like Advancing Analytics changing.
We're not interested in trying to win every role inside a customer's organisation. The better partner model is to find the places where we can add disproportionate value, bring deep technical expertise and industry thinking into the customer team, and then work alongside their people to solve the problem together.
That last part is particularly important to me because the objective shouldn't be to create dependency. Our customer's people know their organisation and industry; we bring specialist technical experience and the perspective that comes from solving similar problems elsewhere. When you combine those two things properly, you create something much more valuable than either team could create in isolation. It also means their internal capability improves as we work, and that knowledge transfer and enablement is almost as important as the actual technology.
That brings us to the question I think most technology and business leaders are wrestling with as they look towards 2027: where should we actually be putting our effort?
Andrew's answer was refreshingly straightforward. He spoke about finding the big-ticket problems that, if you could solve them, would make the biggest difference to the organisation. Then put a strong team around them, make sure they have senior buy-in, and the space to solve them.
He also made another point that I think should be written above the door of quite a few transformation programmes:
“Foundations are key... but if you're building the foundations without the vision, where do you go next?” – Andrew Phethean
And I couldn’t agree more. You don't modernise your data estate simply because modernising a data estate is a good thing to do. You do it because there are important business outcomes that the organisation wants to achieve and the existing foundations are preventing you from getting there.
This is especially true with agents - every big technology event I’ve been to recently is talking about agents, whether that's agents for internal use or agentic commerce across retail, banking and other industries. And that's incredibly exciting, but we also have to be honest about where individual organisations are starting from.
If you have fragmented data sitting across legacy environments, if the business doesn't trust it, if you don't have the governance around it and you can't make that data available to the AI systems you're building, then the sophistication of the model sitting on top doesn't really matter.
“Without that solid data infrastructure, everything else kind of won't work.” – Terry McCann
The last couple of years have been spent proving that AI can do extraordinary things. The next few years are about proving that organisations can do extraordinary things with it. So the question isn’t “what can we build with AI?”, it’s “what can we change with AI?”. Start with the business problem and the outcome, then the platform, the partner and the data foundations all have a much clearer job to do.
---
Terry and Andrew also recently spoke to Shubhangi Dua on the Don't Panic! It's Just Data podcast - you can watch the episode here.