To Build or to Buy? That is the Question Increasingly for MSPs

AI coding tools are empowering MSPs to do the once unthinkable: create tools they would otherwise have paid a vendor for all by themselves.

I mentioned a few weeks ago that I vibe coded an application for the first time over the holidays. More recently, inspired by the buzz about Anthropic’s Claude Opus 4.5, I coded another, and there are more coming. I wouldn’t say I’ve been “Claude-pilled,” as some call it, but it’s close.

And can you blame me? I’ll use this latest system, after a few finishing touches, to manage Channelholic’s editorial calendar, something I’m doing at present in a jerry-rigged Airtable app that’s functional but not much more. The app Claude made for me, by contrast, looks and works like commercial-grade software, is entirely tailored to my workflows, and has taken me maybe two hours to build so far.

Unlike Airtable, moreover, it costs me a whopping $0 a month to use.

Keep in mind that I’m not exaggerating when I say I couldn’t write a line of programming in any language you name to save my life. I really don’t know how to code. Yet that really no longer matters.

“That’s the different world we’re all starting to live in right now, where any of us can just go type into a website what we think we need from a tool, and it can create an 80% structure of that tool for us,” says Joshua Skeens (pictured), CEO of managed services heavyweight Logically. “I think you’ll see a lot more of that.”

So do I, and if we’re both right the implications for MSP software vendors are profound.

“That buy versus build equation has fundamentally turned upside down,” observes Kevin Lancaster, CEO of BetterTracker and Channel Program, during a recent podcast interview. Historically, it’s pretty much always made sense to buy solutions given how much time and money building them would consume.

“Today, it’s almost parity in a lot of cases,” Lancaster says. “It’s probably in some ways more cost-effective to build some of these newer capabilities because of the tools that are out there.”

Now realistically, no MSP is going to ask an employee with zero development experience to vibe code a mission-critical business application. But the same tools that have made my software projects possible will make building solutions you’d otherwise have to buy a real option for someone with some programming know-how.

That’s already happening, in fact, at Logically, where engineers are using AI coding tools to create AI automation solutions, rather than license software from startups still perfecting their products.

“We can build some of this technology in ways that help us today, rather than trying to help another company figure it out,” Skeens says.

AI SOC technology is an example, he continues. “Almost every one of what we consider L1 alerts from a SOC now is completely automated and looked at through automation and AI. Our team barely even looks at it,” Skeens says. “There was nothing off the shelf that we could go buy that would do that for us.”

Glenn Mathis, CEO of mega MSP Integris, came to a similar conclusion some time back when he began investigating AI automation vendors. “They typically are building for the lowest common denominator,” he says, noting that at nearly $300 million top line, Integris has very different requirements.

“The vast majority of MSPs in this country are not platforms. They’re really small MSPs, and the problems that they have with their customers or the scale at which they built their business is very different from mine,” Mathis says.

Which is why Integris has created an AI center of excellence tasked with identifying internal processes ripe for automation. “We’ve invested lots of our resources into this,” Mathis says, noting that it’s getting a lot of return on that investment too. “It saved us dozens of thousands of hours just last year alone.”

In part, he adds, because the software Integris builds meets the firm’s exact needs. “I’m not going to say there aren’t things we will buy and there aren’t things we will partner with, but a lot of things are going to be built by us because we know our customer better than anyone else,” Mathis explains.

Neither Mathis nor Skeens, it’s worth noting, expects to build every system it uses someday. “Over time, it just costs too much money to keep those tools updated and running,” Skeens observes. But some amount of money that might otherwise have flowed to outside vendors will now flow to his dev team instead.

Also worth noting: It will be hard for a lot of MSPs to do the same. “I have 900 people, almost, in my business,” Mathis says. “I can certainly afford to build a department around this.” But if smaller MSPs don’t want to get flattened by the AI flywheel, he adds, they should at least be thinking about how much buy they can replace with build going forward.

“I don’t think an MSP can avoid it,” Mathis says. “You have to dedicate resources to this.”

Shameless plug alert!

Skeens will be joining me and Daniel Gonzalez, head of technology and AI at Shield Technology Partners, for a panel session I’m moderating on “DIY AI” for managed service providers at the Channel Partners Conference & Expo/MSP Summit in Las Vegas this April. Details here. Use promo code “FREEMAN” at registration for discounted entry.