Are MSPs Being Left Behind by AI?

Something's happening in the channel that nobody wants to talk about. While the rest of the tech world races ahead with AI, managed service providers are stuck in a strange limbo. Vendors are promising the moon. Clients are asking questions their MSPs can't answer. And the gap between what AI can do and what most partners can actually deliver is getting wider by the week.

I've been speaking to MSPs across the UK for months now. The frustration is real. One owner told me his team spends more time explaining why they're not using AI than actually solving client problems. Another said she's watching competitors make wild claims about AI capabilities while she struggles to find a single tool that actually works as advertised.

The thing is, this isn't about being slow or resistant to change. Most MSPs I know are desperate to get AI right. They just can't find a way in that doesn't feel like jumping off a cliff.

The gap that's actually forming

Here's what's really going on. While enterprise IT teams have dedicated resources to experiment with AI, test tools, and build internal expertise, most MSPs are running flat out just keeping the lights on for their existing clients. There's no slack in the system. No budget for experiments that might not pay off. No time to become AI experts while also being security experts, cloud experts, and everything-else experts.

The stats back this up. Around 77% of IT staff time at smaller firms goes to support and maintenance. That leaves precious little for anything new, let alone something as complex as AI adoption.

And the vendors? They're not helping. Every RMM platform, every PSA tool, every security product now has "AI-powered" somewhere in the marketing. But when you dig in, it's often just rebranded automation or a chatbot that makes things worse. One MSP told me their vendor's new AI ticketing system created more work, not less. False positives everywhere. Alerts that meant nothing. Time wasted chasing ghosts.

What the vendors aren't telling you

There's a darker side to this AI push that nobody in vendor land wants to discuss. Some of the biggest names in MSP software have been quietly laying off support staff and replacing them with AI systems. Hundreds of jobs gone in some cases. And the partners relying on that support? They're noticing the difference.

Response times are up. Resolution quality is down. The human expertise that used to be a phone call away has been replaced by a bot that doesn't understand context, doesn't know your environment, and definitely can't think creatively about a problem it's never seen before.

This creates a nasty situation. MSPs are being told they need AI to stay competitive. But the AI their vendors are actually deploying is making their own service worse. It's not a good look.

The real risks nobody's talking about

Beyond the hype and the vendor games, there are genuine security concerns that MSPs should be worried about. AI meeting assistants are spreading through organisations like malware. One person in a Zoom call uses an AI notetaker, and suddenly everyone in that meeting gets emails trying to install the same tool. It hooks into M365, gets access to calendars, meetings, contacts. Before you know it, client data is flowing to servers you've never heard of.

HIPAA? GDPR? The AI doesn't care. And neither do most of the companies building these tools.

Then there's shadow AI. Clients are signing up for AI services without telling their MSP. Uploading sensitive documents to public language models. Connecting AI tools to their business email. One wrong click and you've got a data breach that nobody saw coming.

MSPs are supposed to be the trusted advisors here. But how can you advise on something that's moving so fast and changing so often that even the experts can't keep up?

What MSPs can actually do about this

Right, enough doom and gloom. What can partners actually do to close this gap without losing their minds or their shirts?

1. Lock down app consent now

If you haven't already, disable user consent for Entra ID applications across all your client tenants. This stops the AI notetaker worm problem dead. Users can still request apps, but IT gets to approve them first. It's not exciting, but it works.

2. Pick one thing and get good at it

Don't try to become an AI company overnight. Pick one specific use case where AI could help your clients and focus there. Maybe it's security alerting. Maybe it's ticket triage. Maybe it's helping clients write better emails. Whatever it is, learn it properly before moving on. Depth beats breadth right now.

3. Have the honest conversation with clients

Your clients are being bombarded with AI marketing from every direction. They need someone to cut through the noise and tell them what's actually worth their time and money. Be that person. Even if the answer is "most of this stuff isn't ready yet," that's valuable advice. It builds trust.

Where we go from here

The AI gap is real, and it's not going away. But MSPs have been here before. Remember cloud? Remember SaaS? Remember when everyone said managed services would kill break-fix? The partners who survived those shifts did it by staying practical, staying honest with clients, and not chasing every shiny object that vendors waved in front of them.

AI will be the same. Some of what's being promised today will become essential. A lot of it will quietly disappear. The MSPs who come out ahead will be the ones who took the time to understand what actually works, what their clients actually need, and what risks are worth taking.

It's not about being first. It's about being right. And that's something MSPs have always been good at, when they give themselves permission to slow down and think.

The gap is there. But it doesn't have to stay that way.