9 hours ago
2
ARTICLE AD BOX
It’s taken a while, but most organisations have finally accepted that their future resides, at least in part, in the cloud. Now all they have to do is get their heads around AI. But their biggest challenge in making that transition is unlikely to be technological, argues Microsoft UK veteran Alex Grimshaw.
Grimshaw’s career has taken him from management accountant at the Co-op to Microsoft UK’s chief technology officer for central government and, for the last five years, CTO for UK Azure. That means he’s become intimately equated with the problems that massive migrations can inflict on thousands of users and millions of customers. And yet, he says, the biggest barriers to successful transitions are invariably cultural.
In the following interview, edited for length and clarity, Grimshaw tells Tech Monitor about technical debt, who really influences a transformation in the enterprise and why some customers are wasting their money on AI initiatives.
Cloud migration is full of knotty, controversial and plain difficult cases. They’re also the most interesting to make happen, says Microsoft UK’s Alex Grimshaw. (Photo: Microsoft)Your career has straddled the line between the public and private sectors. What did you learn working on both sides of that divide?
That some organisations have a habit of hyper-focusing on technical debt, or debt in terms of their legacy infrastructure, whether it be Microsoft’s stack, VMware, IBM mainframes, or even COBOL. I still find the main success factor working with those organisations is attacking the cultural debt and the huge reliance on the outsourcing incumbents.
You cannot just rock up at the end of something and go, “there’s a new system”. If you don’t tackle the change management, really, it’s not going to work.
What was the project you’re most proud of?
When we moved [HMRC’s] SAP-based enterprise tax management platform. That platform collects over £365bn of tax revenue every year. If someone said to me, ‘Describe the characteristics of a mission-critical, critical national infrastructure type project,’ it would be that. You can choose the most gnarly, interdependent workload you can find if you’re experimenting or you can dip your toe in there and pick something really easy. So, we jointly decided to go for the biggest, hairy, scariest one we could find.
You say that it’s important to connect with all the stakeholders involved in a given cloud migration, not just the decision-makers. But when it comes to an organisation like HMRC or DWP, how does a technology leader connect with all the people involved?
I always come back to the ‘why’ of the project – who cares about why this is ultimately happening?
We can’t go out shouting from the rooftops how amazing this transformation is going to be. We’ve just got to focus on finding the right senior advocates, the right champions, and get them on board, because then, hopefully, you’ll have that cascade effect that goes down the organisation.
[Think about] what it means for me as a user of that system. It’s dying often, or it’s a bit slow, or I’ve got to copy and paste from this system to this system, all those kinds of things. From a vendor perspective, you’ve just got to think about the different personas in terms of stakeholders and what it means for them.
How have board-level conversations with customers changed compared to 10 or 15 years ago?
What I’ve found, over the last 12 months, talking to customers certainly from a board level, is that they’re always asking, “How do we adopt this AI thing?”
And I’ve seen some very odd behaviours triggered by this “let’s do AI” mentality. So, of course, that then trickles down to IT, and you see behaviours like replacing chatbots with AI-type chat functionality. Well, that’s not embracing AI. That is simply spending more money on a service that does the same job as your chatbot did.
Then, alternatively, you are seeing some wholesale “We want to embrace AI,” which means I’m now seeing more customers creating or reassessing their entire data strategy – one that might already be a really robust data strategy that has governance in there and consistency and quality and availability. You cannot really build an AI layer on top of something unless it’s robust from a data perspective.
It’s hard to separate AI from questions of data and digital sovereignty. How do you have that conversation with customers or potential customers?
Whether it’s digital sovereignty, data residency or data sovereignty, it’s all about defining your compliance requirements.
I think it’s wrong of me or Microsoft to label [the customer’s] requirements. Ultimately, it’s their requirements. So, it’s about us understanding what those are.
I think customers are looking more at what digital sovereignty means to them. Where do they need to hold their data? Where do they need to consume it from? And as important as that, where do their customers need the data to [reside]. We’ve got to think about that as well.
How is Microsoft beginning to learn more about the potential of AI products from internal use cases?
For Copilot, we are, of course, customer zero. If we cannot adopt it internally, then of course, how do we expect customers to? So, we’re going through lots of our own learning journeys to understand how customers could adopt this stuff in the same way that we’re speedily trying to get value from it as well.
I’ll give you an example I came across not long ago. I didn’t appreciate that a lot of people who have autism or ADHD suffer a lot of anxiety and stress when they don’t understand the tone within an email or comms. So, what they’ve been doing is utilising Copilot. So, that email from their manager, they can see the actions. They see the black and white binary bit. But what’s the tone? What’s the intent? Is their manager angry with them or happy with them? So, they ask Copilot ‘What’s the tone? How do you feel? How do you think my manager’s feeling right in this email?” So, it then reassures them that the use of language is a neutral or positive one. I’m just mesmerised, even by that one use case.
