The next evolution of managed services may require MSPs to rethink what they actually sell.
As artificial intelligence becomes part of everyday business operations, managed service providers have an opportunity to move beyond maintaining IT infrastructure and become the advisors customers rely on to implement, manage, and improve AI-powered workflows.
That’s the emerging role of the AI managed service provider, or AIMSP.
For MSPs, the transition creates opportunities to build consulting, implementation, and recurring services around AI adoption. But it also demands a different kind of customer relationship—one built on continuous innovation rather than simply keeping technology running.
In the first installment of The AI MSP, Howard M. Cohen explores what this transition means for providers and introduces four levels of AI adoption that MSPs can use to develop new services.
Why AI Changes the MSP-Customer Relationship
You will become a tremendous enabler as your customer’s AIMSP.
It is likely that many of your customers’ employees use AI to some extent. It is unlikely, however, that they have gained sufficient expertise to be relied upon to support the business they work for. Most consumers never get beyond the most fundamental uses of AI, as we’ll discuss shortly.
Everything you have ever sold to any customer has been sold with the intent of enabling them to do things better, more efficiently, to do things less expensively, and simply to do more things.
Most of that enablement was foundational. You built the platform on which applications could run. You kept them running.
AI is there to do all the things you don’t want to have to do yourself.
What makes anything into something you don’t want to do?
For example, you are preparing a report, and you realize you need to look something up. You could open a search engine and start looking for websites where you might find the information you need.
You know you’ll have to create documents and ultimately consolidate your information into something that can be easily digested, but you’d rather spend your time creating the analysis around the report, not the basic data itself.
The picture quickly becomes clear. AI has access to more information than has ever been assembled before, and it can find everything incredibly quickly. Give it what you’re looking for, and it will find it among hundreds or thousands of documents.
What Your Customer May Not Yet Know But You Will
First and foremost, AI is a toolset. It provides you with a wide variety of very capable tools.
They say a tool is only as good as the craftsperson who wields it, which suggests you need to learn how to use them.
Not AI.
To use any AI tool, you simply ask the AI chatbot how to use it. Or you ask that chatbot to do it for you. You tell it what you need done, in what kind of setting or context, with what barriers and obstacles, and some idea of how you’d like it delivered to you. Often, the AI gets it right on the first try.
Sometimes it won’t. But so what?
You’ve heard lots of names of skills tossed around like “prompt engineering” or “context engineering” and more. You’ve heard about “vibe coding.”
Forget it. We’re beyond all that! What you need to do to work effectively with any AI model is simply called “conversation.”
And it’s okay if the AI model gets it wrong!
If the AI returns results that don’t exactly match what you were looking for, the conversation continues. You simply tell the AI what it got wrong. Give it as much detail as you can.
It’s called “iteration.” You and your AI simply keep conversing back and forth, improving the output with each exchange until it’s exactly what you want. Not unlike collaborating with another person, you keep on talking about it until you get it right.
What You’ll Need to Embrace
You’ve been down this road before. You install a new system. You train the users to use it. Then you train them to support it while you support them. At some point, they no longer need your support. You have made yourself obsolete in their environment.
That’s exactly what will happen with AI. So, what do you do about it?
You need to embrace the idea that you will need to keep innovating and adding new features, processes, and ideas to bring to your customers that they will want to adopt to improve their business.
What is Your Relationship with Your Customer?
Done right, you become your customer’s resident AI expert. You’re the one who keeps up with all the new innovations being introduced at breakneck speed. You’re the one who guides them to those who serve them and away from those who don’t. You meet with them to strategize ways to revolutionize their business with these new AI advances.
You become their key competitive advantage.
What Comes Next?
Over the next four columns, we’re going to explain the four levels of AI that you’ll be guiding your customers through.
If your customer sees AI as this huge challenge, we’ll break that challenge down into smaller, more digestible bites.
If your customer fears AI, we’ll show them why they have little to fear.
If your customer doubts they can even get started with AI, we’ll show you how to smooth the path for them to become familiar with it by starting slowly and only working our way up when they’re ready. Hint: it’s very possible 80% of them will never get past level one.
Four Levels of AI Services MSPs Can Offer
We’ll break AI down into four levels:
The Assistive Level
The AI exists to assist you by doing the things you really don’t want to do yourself. We’ll describe many of the most useful things they assist with. 80% of users may never need more than this!
The Copilot Level
As the things you want to do get more sophisticated, you’ll want your AI to collaborate with you, show you how to do certain tasks, and handle others for you. Microsoft called their first AI product Copilot for just this reason.
The Agentic Level
AI can take on very sophisticated processes and procedures for you. You might say “it takes a village” for them to achieve truly sophisticated results. That village consists of a “team” of AI “agents,” each assigned to a specific job, with one serving as the supervisor of the others. We talk about keeping a “human-in-the-loop,” so the supervisor agent can report their progress frequently to ensure projects stay on track.
The Fully Autonomous Layer
Here’s the layer people worry about. The agents are built to perform their tasks on an ongoing basis with little or no human supervision. Only a small percentage of AI projects built worldwide are operating at this level, mainly because it’s difficult to identify which use cases require this kind of operation.
Once we’ve gone through all four levels of AI, you’ll have a comprehensive set of services to sell. We’re going to talk about what you can help your customers do at each of these levels, and how you can build your services at these levels so that you can go sell them, deliver them, and provide ongoing support for them.
What’s Next in The AI MSP Series
While AIMSP will publish twice monthly, you can also consult AgenticMSP, a Substack that delivers “all the information an MSP needs to build their AI Empire.”
Each issue we’ll talk about strategies, profile experts, discuss new product introductions, show how to best consult to your customers about AI, and more.
I also share my homework! For each issue, we review thousands of recent articles to select a group that will be very relevant and of great interest to MSPs about AI. For each article, you’ll get a summary, an explanation of why it’s so MSP-relevant, an Action List so you can take what you learn and put it to use, and the link to the original article.
The goal is to eliminate the wasted time spent looking for pertinent info so you can focus on the news that is worth your time.
Please reach me at hmc@howardmcohen.com with any questions you may have about AI resources and how MSPs build their AI business.
And follow Channel Insider’s coverage as we break down AI and all of the other technologies MSPs still need to manage for their customers.