AI governance is emerging as a recurring managed services opportunity for MSPs, as customers struggle to maintain continuous control over permissions, stale content, and other risks across Microsoft 365 environments.
ShareGate, a Microsoft 365 management and governance software provider, argues that governance should no longer be treated as a one-time prerequisite for deploying tools such as Microsoft Copilot.
Richard Harbridge, principal industry advisor at ShareGate, told Channel Insider that 78% of AI leaders surveyed by the company said governance activities—including permission reviews, cleanup, and lifecycle management—directly affect their confidence in investing further in AI.
“Increasingly, we’re seeing it as an enabler of AI, rather than simply a control around it,” Harbridge said. “It’s the way to safely accelerate.”
AI readiness requires continuous governance
That shift has implications for partners selling Microsoft 365 and Copilot readiness services.
“I wouldn’t describe Copilot readiness, and I think a lot of people do, as data cleanup, as a checklist,” Harbridge said. “Because that suggests you can be data ready, and you’re checked, and you’re good.”
Instead, organizations need the ability to continuously monitor their environments, make governance decisions, address problems, and demonstrate that their controls remain effective.
“Readiness never ends,” Harbridge said.
That continuous requirement is where Harbridge sees an opportunity for partners. While an AI deployment might be handled as a discrete project, he said governance “has to be this operating discipline,” and many customers lack the resources or expertise to maintain it internally.
AI accelerates existing Microsoft 365 risks
The underlying challenges, including oversharing, outdated permissions, abandoned workspaces and stale information, largely predate generative AI. What has changed is the speed at which those problems can produce consequences.
“AI turns governance debt into business consequences faster than before,” Harbridge said.
ShareGate’s research found 71% of AI leaders said their governance workload increased after enabling AI. Harbridge also pointed to a striking disconnect between confidence and outcomes: 96% of respondents expressed high confidence in their Microsoft 365 governance posture, while 29% reported that AI exposed content it should not have.
Partners should avoid fear-based AI selling
For partners, Harbridge cautioned against turning those risks into a fear-based AI sales pitch.
“The trusted advisor isn’t the partner who makes AI sound dangerous,” Harbridge said. “It’s a partner who can show where the real risk is and what to do about it” on a continual basis.
As SMB, midmarket, and enterprise customers continue to adopt AI, partners are expected to help with far more than traditional tech implementation and services asks.
AI agents introduce delegated authority risks
That need could become more acute as businesses move from copilots that surface information to AI agents capable of taking actions across enterprise systems.
“We’ve spent a lot of time with Copilot asking, ‘What information can AI see?’” Harbridge said. “With agents, increasingly, we’re talking about delegated authority.”
The questions then become what an agent is doing, which systems it can act within, whose authority it carries, and what limits govern its behavior.
Governance could become a recurring managed service
For MSPs, that changes the conversation about managed services.
“The next managed service question is less, ‘Is the system online?’” Harbridge said. Instead, partners will increasingly need to determine whether AI actors — effectively new digital identities — “are operating within the authority we intended.”
And that makes solving today’s governance problems more urgent.
“Imagine how much worse that is when an agent does that at a speed and scale agents can do,” Harbridge said of persistent oversharing and access problems.
For partners, the opportunity is therefore not just preparing customers for Copilot. It is helping them build a governance model capable of keeping pace with what comes after it.