MSP 3.0: why AI adoption, not break-fix, is the next managed service.

The managed-services business has reinvented itself twice in twenty years. A third reinvention is arriving now, and it has nothing to do with patching servers, it's about helping customers adopt AI without getting hurt.

Two reinventions, and a third arriving

MSP 1.0 was break-fix: something broke, you fixed it, you billed for the hours. MSP 2.0 was the managed-services and cloud era, recurring contracts, monitoring, Microsoft 365 administration, security as a line item. It rewarded providers who moved from reactive labour to predictable, productised services.

MSP 3.0 is the AI adoption partner. Every customer wants AI in their business. Almost none of them can deploy it with confidence on their own. The provider who can make AI safe to switch on, and keep it safe, becomes indispensable in a way that a help desk never was.

Customers aren't buying AI. They're buying confidence.

This is the shift that matters. A customer doesn't actually want "Copilot" or "an agent", they want the outcome those tools promise, delivered in a way that won't expose their data, breach a regulation, or blow up in an audit. They are buying confidence that the technology will behave. That confidence is a service, and it's one a trusted provider is far better placed to deliver than a software vendor.

The blocker isn't the model, it's identity

When a customer switches on AI in Microsoft 365, they don't just add features. They create identities: Copilot agents, custom agents, service principals, OAuth-connected apps, each with standing access to mailboxes, files and data. Nobody can easily answer the basic questions: what exists, what can it reach, who owns it, and is it still needed? Until those questions have answers, "yes" to AI feels like a risk, and projects stall in pilot.

That governance gap is exactly the problem an MSP is positioned to solve, and to bill for.

Why the MSP owns this

Three reasons. First, you already run the tenant: you have the access, the relationship and the day-2 engagement that a vendor doesn't. Second, you operate at scale, the average MSP supports around 120 client organisations, most of them on Microsoft 365, so a productised AI-governance service multiplies across a book you already own. Third, trust: when a customer asks "is this safe?", they ask you, not the software.

What the service line looks like

• Assess, free. Connect a tenant read-only and produce an NHI risk picture the same day, the discovery artefact that starts the conversation.
• Onboard safely. A paid AI Identity Onboarding Review that turns findings into a remediated, governed starting point.
• Govern continuously. Identity sprawl never stops, so monitoring and quarterly posture reviews keep you embedded, a recurring service, not a one-off project.

That is the shape of MSP 3.0: lead with a free assessment, land a safe-onboarding engagement, and convert it into recurring governance revenue, on the clients you already serve.

Industry context: average clients per MSP from Datto, Global State of the MSP Report (2024). Sabiki is not affiliated with or endorsed by Datto. Source.