Becoming the MSP of the future: why we made ourselves our own first AI customer

AI is the biggest new capability to reach business in a generation. For our clients it isn’t an upgrade to something they already had. It’s a genuinely new capability that can read, reason, draft, analyse and, increasingly, act. The benefits are real and they’re measurable. It drives revenue, it creates capacity, and it opens up products and services that weren’t possible before. Clients can see the upside, and they want it.

But the same power that makes AI valuable is what makes it costly to get wrong. This isn’t ordinary business software. A mishandled model can leak data, mislead a decision, or put something indefensible in front of a client or regulator. Even governments that are actively championing AI are now tightening security reviews and cracking down on AI-enabled attacks. The opportunity is real, but so is the risk of getting it wrong.

That’s the line we’ve chosen to walk. We’re becoming the MSP of the future, moving from value-added reseller, to managed service provider, to a managed intelligence partner. Someone who helps you put that intelligence to work. And we decided we had no right to help a client adopt AI until we’d properly adopted it ourselves first.

That discipline has a name inside our business: Customer Zero. We use our own people, our own data and our own go-to-market activity as the first proving ground for every tool before it reaches a client, not as a marketing line, but as an operating principle. It means our consultants advise from lived experience rather than a slide, because they have navigated the awkward middle themselves: the first false starts, the governance questions, the moment a team stops being wary of a tool and starts depending on it. And it means we have already worked out how to do it safely, human-led, ethically, securely and with proportionate governance, before a client’s data or reputation is ever on the line.

In this article we share those key learnings openly to help our clients on their AI journey.

Give adoption time to stick

The first thing being Customer Zero taught us is that the hard part of AI is not the technology, it is people. Equipping our team with Microsoft Copilot and Claude was the easy part; turning that into genuine, confident, safe daily use was the work. AI adoption is behaviour change, and behaviour change needs practice, reinforcement and time to embed, not a single burst of enthusiasm that fades by the following month.

So we built a structured, human-centred adoption approach and ran our own people through it first. It starts by meeting people where they actually are, measuring readiness across dimensions like confidence, perceived impact on their role, trust in leadership and openness to change, because someone quietly anxious about being replaced does not respond to the same intervention as someone already experimenting. From there it moves through role-specific use cases, prompt and pattern playbooks, workflow redesign, and a measurement layer that tracks behaviours rather than course completions. It also surfaces the champions, the early adopters who pull everyone else along.

That same programme is now a client service, because the problem is universal: capability does not spread through announcements. What we learned mapping our own workforce, where the confidence gaps are, where the trust gaps are, which roles unlock the most value, is exactly what clients need before they spend a penny on training. It lets them target support and avoid the most common waste in this market: blanket programmes that treat a finance analyst and a frontline manager as if they need the same thing.

Why we built Pathfinder

The second lesson came from the commercial side of our own business. We needed a better way to identify and sequence transformation opportunities across our client base, so we built one, and became its Customer Zero too. Pathfinder is our outcome-led, agentic methodology. It ingests real signals, conversations, multi-source data, and builds a prioritised roadmap of Microsoft investment across Modern Workplace, Azure and Fabric, with security and governance built in from day one, then turns those foundations into repeatable AI solutions we can scale.

The insight underneath Pathfinder now anchors every client conversation: the prerequisites for agentic AI are no longer separate workstreams. Azure AI Foundry and Microsoft Fabric have converged into a genuine platform for agentic solutions. Modern workplace, cloud, data, security and AI are not five projects, they are one investment roadmap. The organisations that build those foundations now, with governance baked in, are the ones that will be agentic-ready when the opportunity arrives rather than scrambling when it does. Having proven Pathfinder on ourselves, we are now extending the same capability to clients. That’s the natural arc of Customer Zero.

Agents already at work in our business

Customer Zero isn’t theoretical, and it isn’t finished. It’s happening now. Agentic workflows are already running inside BCN’s own operations, in live use rather than in a lab, with more being built as we go.

We are deliberately specific about this, because it is the entire point of Customer Zero: when we tell a client what an agent could do for them, we are describing something we are already running on ourselves, including everything we learned making it work safely.

Build on what you already own

There is a temptation in AI to start from a blank sheet. We think that is usually a mistake, and a slow, expensive one. Most organisations, especially in the mid-market, are already sitting on significant Microsoft investment they have never fully used: identity they have paid for, data scattered across systems that were never connected, licensing that includes capabilities nobody has switched on, workflows that could be automated tomorrow. The fastest, lowest-risk route to value is not to replace it but to activate it.

Our role is to be the glue. We combine platforms, data and workflow into solutions that work for organisations too large for off-the-shelf tools but too small to build everything themselves, starting from the Microsoft estate the client already owns and turning it into agentic-ready foundations. It is faster and lower-risk than rip-and-replace, it respects the money already spent, and because it builds on what is already in place, the value shows up sooner.

Microsoft-anchored, deliberately multi-model

Being Microsoft-centric doesn’t mean being single-vendor. If anything, the AI era has made clear that no single provider has all the answers. Not every job needs the same AI behind it, and what should decide that is the sensitivity of the data, not whatever’s in fashion this month.

So in practice, Microsoft is our anchor. It’s the governance and productivity layer that our enterprise and public-sector clients trust and, in many cases, are required to use. From there we pick the right engine for the job, whether that’s Microsoft, Google or OpenAI, based on cost, capability and what we’re actually trying to do. We keep a close eye on the frontier models too, because new ones land faster than most governance can keep pace with. Our advice to clients doesn’t change: advisory first, build second. Ask what the data is, how sensitive it is, what outcome you’re after, which tool fits and how you’ll govern it. Then build.

Why we’d rather get it right than get it fast

This is where Customer Zero matters most. The reason we prove AI on ourselves is that the technology is powerful enough to demand it. For every use case we ask the same question: not how AI can be applied quickly, but how it can be applied appropriately.

In practice that means treating governance as a living capability, not a one-off policy document. There are clear boundaries on where AI can be used freely, where it can assist but a human must own the final output, and where content must be human-led entirely: client-facing communications, regulated statements, anything carrying reputational or legal weight. It means GDPR-compliant gateways, anonymised data and proportionate guardrails, and keeping activity, wherever possible, inside the Microsoft 365 environment the client already governs. For regulated and public-sector clients it also means respecting data sovereignty and the transparency obligations they answer to. Because we have run all of this on ourselves first, we can give clients something rare in this market: not just a way to adopt AI quickly, but the confidence that it is being done safely.

That’s what Customer Zero is ultimately for. It isn’t about us. It’s the discipline of proving value, and proving safety, on ourselves first. So that when we bring AI to a client we are not selling a promise, and we are not asking them to take a risk we have not taken ourselves. We are sharing something that already works, and that we already know how to govern.