In May we hosted our latest Cloud @ WRK roundtable at the Bank of England, focused on a topic that’s quickly moving up every organisation’s priority list:

👉 How do you actually prepare your cloud environment for AI?

A big thank you to everyone who came along and contributed — the quality of conversation in the room made this one particularly valuable.

 

A shift in the conversation

What was interesting from the outset is how much the discussion has evolved.

Not long ago, cloud conversations were centred around migration, cost, and platform choice.

Now, the lens is very different.

Cloud is increasingly being seen as the foundation for AI, not the end goal.

And that shift is forcing organisations to ask some important questions:

• Are we set up to safely scale AI?
• Do we understand the risks behind what we’re building?
• Can we track value — not just cost?
• Who actually owns these decisions?

 

The role of the Cloud CoE is changing

One of the strongest themes throughout the session was how the Cloud Centre of Excellence needs to evolve.

Historically, many CoEs have focused on governance, standards, and cost control — but that model alone doesn’t hold up in an AI-driven world.

What’s emerging instead is something more dynamic:

👉 A function that doesn’t just protect the organisation
👉 But actively enables teams to move faster — safely

It’s less about saying “no”
And more about creating the conditions to say “yes” with confidence

 

It’s not one problem — it’s four, all at once

Another key takeaway was how intertwined everything has become.

You can’t talk about AI in the cloud without simultaneously considering:

• Governance
• Security
• Cost
• Sustainability

The challenge isn’t solving one of these in isolation…

👉 It’s designing an operating model that brings them together

 

What does “good” actually look like?

The most valuable part of the discussion was moving from theory into how this works in practice.

A few patterns that stood out:

• Clear intake processes for AI use cases
• Defined ownership across models, prompts, and outcomes
• Built-in guardrails from day one
• Real visibility of cost at use-case level
• A structured path from sandbox → production

And importantly — starting small, but with intent.

 

The reality check

There was also a lot of honesty in the room around where organisations are today.

Many are still:

• Experimenting in pockets
• Lacking consistent governance
• Struggling with cost visibility
• Unsure how to scale beyond pilots

Which makes these conversations even more valuable — because while there’s no single blueprint, there are clearly shared challenges.

 

From a hiring perspective…

This is already shaping the market.

The conversations I’m having are less about pure cloud engineering, and more about people who can:

• Connect cloud, AI, and business outcomes
• Balance speed with control
• Navigate cost, risk, and performance
• Enable teams — not slow them down

 

Final thoughts

If there was one way to sum up where organisations are right now, it’s this:

👉 AI doesn’t fix your cloud — it exposes it

• If your foundations are strong, AI will accelerate you
• If they’re not, it will amplify the gaps just as quickly

There was also a clear sense that while multi-cloud is a reality, focus still matters:

👉 Start with a primary cloud, build properly — but design with enough flexibility to move where needed

And underpinning all of it:

👉 The basics matter more than ever

Because when you start layering in AI, things like architecture, governance, cost control, and security don’t get replaced…

They get pushed to their limits.

 

If you couldn’t make it and want to catch up on the themes or grab a copy of the slides, feel free to drop me a message 👍

And as always — appreciate everyone who continues to support and contribute to the Cloud @ WRK community 👏