Why the MSPs that win the AI shift will look nothing like the help desks we all grew up on. 

By Jimmy Hatzell, CEO and Co-Founder, Hatz 

I've seen this movie before. 

In 2018 and 2019, I was at IT Nation with a lot of other security people, all looking at each other and telling MSPs the same thing: you are going to need more security. We were right. It just took a few years. 

AI feels like that moment, with one big difference. The clock. I think this wave is moving about six times faster than the security wave did. 

Here is the other difference. Security is a spend. Customers hate paying for it and hope they never need it. AI is a competitive differentiator. It drives costs down and output up. That is why I have said for a while that AI is a bigger opportunity than security, and why I think the average small business will spend more on AI than on cybersecurity within two years. 

At the beginning of 2026, a customer might have told you, "We tried AI." Now it is hard to find one who has never used an LLM. The demand is like nothing I have ever seen. 

So the question is no longer whether your customers will use AI. It is who is going to help them. 

AI is not a chatbot, and it is not a seat 

When people hear "AI," they picture a chatbot. It is not a chatbot. It is a tool you can tailor to how a specific business actually works, and that is when you start to see real business outcomes. 

At Hatz, we call what we are really building digital labor. It is where we started. The idea is simple: find the work that takes a long time but does not need deep expertise, and hand it to software that can do it. 

What changed is the software. It can now run for longer, repeat the work reliably, and use real judgment along the way. 

This matters for your business model, not just your tech stack. Think about what happens to per-seat pricing when your customers hire fewer people and spend more on intelligence. Per-seat monthly recurring revenue starts to break. 

Digital labor changes what you sell. You are not selling a license or access to a tool. You are selling work that got done. 

Adoption flipped the script 

Cloud. VoIP. Every big shift MSPs have navigated came from the top down. Leadership decided, and you deployed. 

AI shows up from the top and the bottom at the same time. Executives want it. Employees are already using it, sometimes with more expertise on a specific use case than their IT provider has. Business integration is happening before security and controls. 

That is why "nobody is allowed to use AI" is not a real policy. You cannot tell a customer they cannot use AI. They will use it anyway, on personal accounts, and you will not see any of it. 

The job is to keep the utility and take out the risk. We frame it as three controls: 

• Security. Where does the data go? Read-only access, no training on customer data, and clear retention terms. 

• Spend. What is this going to cost? An AI code review tool can run up a $5,000 bill in 24 hours if nobody is watching. 

• Scope. Who is using it, and for what? 

And if you are wondering where to start, start with shadow AI. It is already happening inside every customer you have, which makes it the easiest first conversation. 

The advantage MSPs already have 

Most AI consultants have to build three things from scratch: trust, proximity, and knowledge of how the client operates. You already have all three. 

I think of MSPs as forward-deployed engineers. You stay close to the customer, and you translate raw AI capability into workflows that solve real problems. Think about who has shown up every time a small business needed help with technology. It has been MSPs. 

There is a gap to close. Only about 30% of small businesses are experimenting with AI, compared with more than 70% of large enterprises. It is not a lack of interest. Most of them do not know where to start. 

The MSPs making the most progress are not selling AI as an add-on. They are weaving it into marketing, sales, support, and technical operations, and they are selling outcomes: less repetitive work, faster response times, fewer manual steps. 

AI is a show-don't-tell technology 

Here is the problem I hear most. A customer says they want AI, and an MSP asks, "For what?" Then everyone stares at a blank screen. 

AI is one of the first technologies where you cannot see the full potential by looking at it. You have to watch somebody use it. 

I learned that in my own office. Rachel on our marketing team makes animated videos in Hatz Artifacts that I didn’t know were possible. She works in marketing. I live in code. Same platform, completely different unlocks. I would never have found hers on my own. 

That is the idea behind Hatz Activate, our go-to-market feature for MSPs. It gives you one place to see where every customer stands and a repeatable way to start the conversation. We are pushing it further, toward helping you spot the use cases and the biggest opportunities for each business. 

You do not tell customers what AI can do. You show them something real, with their data, in their workflow. 

Own it, or it does not happen 

When I talk to MSPs who are making this work, one thing is the same every time. The owner is in it. 

I would go further. The MSP owner, not a new hire, should start the AI practice, with the first two or three customers. This is an existential revenue question for a lot of MSPs over the next two years. It cannot be handed off, because sometimes the rules have to bend for it to succeed. 

The owner also goes first inside their own company. You cannot guide customers through something you have not used yourself.

What it looks like when you run on it 

We are customer number one. We use Hatz to build Hatz and run Hatz. Our engineers each have their own software factory, with agents hunting bugs and reviewing code, and humans still in charge. Our CFO sends a financial model along with an AI-built walkthrough of it. Our research team tests models by asking them to build Tetris. (Yes, I build games in Hatz. It is for science.) 

Running this way teaches us what MSPs will need before they ask. We would not know how to build a secure agent vault for a business if we had not needed one ourselves. 

It also keeps us honest on spend. Smarter model routing cut our own costs by about 75 percent on common workflows. And connecting AI to the tools you already run, like your PSA, documentation, and security stack, is what lets it do real work instead of just answering questions. 

Where I would start this quarter 

1. Run it on yourself first. Put AI to work in your own company before you pitch it. 

2. Open with shadow AI. It is the easiest conversation and the most urgent one. 

3. Pick two or three customers. Show them something real, not a slide. 

4. Set security, spend, and scope before you scale. Control is what lets people use AI more, not less. 

5. Sell the outcome. Hours saved, faster responses, work that no longer piles up. 

6. Make it recurring. AI becomes another managed layer of the customer's environment, with ongoing governance and optimization, not a one-time project. 

Customers are not waiting 

Your customers are no longer waiting for you to figure this out. If you cannot offer the solutions they want, that is a real gap. 

I would rather you fill it with us. Hatz only builds for MSPs, and that is not changing. 

I covered a lot of this in the Hatz October Partner Roundtable. If you missed my talk, watch the recording here