The MCP Acid Test for AI and Community

Spend any time at all in the IT industry and you’ll quickly notice how freely we toss around the word “community.”

There’s the vendor community, for example, a collection of software and services companies prone both to mutually beneficial alliances and cutthroat competition. Then there’s the partner community, whose members so readily share the best of best practices with one another as to leave newcomers scratching their heads in amazement. And then, of course, there’s the channel community, where vendors, partners, and other stakeholders come together in pursuit of the common good.

I spent last week at three conferences in as many cities on two coasts exploring the interplay between AI and community. MCP and its implications for the channel community are a good place to begin discussing what I heard.

In theory, the open exchange of data made possible by MCP is very much in the interest of the channel community. In reality, whether or not the vendor community will see it that way is still in question. Steve Petryschuk, VP of product and market strategy at network management vendor Auvik, thinks it very well might, both because MSPs want it to and because software makers know it.

“As far as I can tell, most of the vendors within the MSP community understand that a rising tide raises all ships, and so if we collaborate we can grow our businesses together,” he says.

On the other hand, he continues, “that’s a very positive outlook on things, and I’m not sure that I’ve always been the most positive person.” Vendors are profit-driven entities with a history of preferencing control over cooperation.

“I think we’ll have to see how it plays out,” Petryschuk says.

So far, so good from what I can see. N-able launched an MCP server in April, and while fellow vendors may one day have to pay for access to Kaseya’s recently introduced MCP server, it will only be enough to help cover infrastructure costs.

Syncro (as savvy Channelholic readers like you now know days ahead of everyone else) will join the club this Tuesday as well in an especially big way based on a similar reading of market demand.

“We believe that MCP and extensibility are going to become a buying criteria,” explains Dee Zepf (pictured), the company’s chief product officer.

Syncro has a particular kind of extensibility in mind, however. To ensure users (and Syncro itself) retain control over their data, the company will provide access to MCP data exclusively through ChatGPT, Claude, Copilot, and the like, which will serve as ersatz integration hubs for entire MSP tool stacks. Say I use Syncro for RMM and PSA and Hudu for documentation.

“I would connect to my Hudu instance from Claude and I would connect to my Syncro instance from Claude,” Zepf says. “You’re not actually connecting Hudu to Syncro. You’re just connecting them both to Claude, and then Claude is stitching them together for you.”

Which raises a second question. Let’s stipulate that everyone wants to consume data via MCP. How do they want to consume it? Syncro believes its partners should make that choice, but its integration strategy implies a prediction for what they’ll choose. MSPs, the company believes, will eventually find the convenience of talking to an LLM on their phone when they need to close a ticket on a Sunday afternoon irresistible.

“You just need a few words of natural language and you’re back to whatever you were doing,” Zepf says. “It’s a new way to work.”

And a better one, Syncro contends, than relying on an embedded chatbot in an RMM or PSA solution. “You’re a little bit at the mercy of the vendor that puts the copilot in their product, how much they want to charge you for it, what its limitations are, what it does, which models it uses behind the scenes, and how often those models get updated,” Zepf says.

Ingram Micro thinks partners want LLM access to data too and introduced an MCP server last week to make that possible. Houston-based MRE Consulting, which helped beta test the system, likes ordering through the distributor’s Xvantage platform, and likes using the Xvantage Integration Hub to transact directly through its HaloPSA implementation even more. “That substantially sped up our quoting process,” says Zach Esmond, MRE’s director of strategic partnerships. The MCP link, though, has further accelerated ordering speed “exponentially,” he adds.

Auvik, whose MCP server is in private beta now and due to go GA in the fall, suspects most MSPs will have the same experience and similarly prefer to use someone else’s interface to connect with its solution. It also isn’t the slightest bit worried about it. Users have been accessing its network data through their PSA system for years.

“If that switches to Claude or to OpenAI or something else, I’m actually not all that concerned for Auvik’s sake,” Petryschuk says. “We’re still going to be that source of truth for the network.”

If anyone’s vote counts in this matter, though, it’s got to be Slide’s, right? The backup vendor introduced its first AI-native MCP interface a year ago, ahead of pretty much any other vendor serving MSPs.

“Since then, a lot of that capability has actually become main code and moved into the product itself,” reports Carlson Choi, the vendor’s COO. And that’s generally where users want it too, he adds. Only about 5-10% of Slide’s roughly 1,000 MSP partners are taking advantage of the option to connect with backup data from inside LLM chatbots, and they tend to be large companies with sophisticated, deeply bespoke tool stacks.

That’s consistent with what Syncro’s seeing too for the moment, according to Zepf, but it won’t be long, she predicts, before large MSPs aren’t the only ones with sophisticated, deeply bespoke tool stacks.

“MCP is making that more approachable for folks downstream,” she says. “It’s like an API 2.0, but it’s easier to use and consume than an API.”

Syncro goes native

It’s a measure of just how deeply convinced Syncro is that MSPs will eventually do most of their RMM and PSA work via LLMs that it’s created a connector for Claude and applied to have it listed in Anthropic’s official directory. And on Tuesday the world will learn something else that Channelholic readers already know, which is that Anthropic has accepted the application.

“We show up as a native connector,” says Syncro CEO Michael George (pictured). As far as I can tell, that’s a first for an RMM/PSA vendor and one that puts Syncro alongside names like Slack, QuickBooks, Docusign, and Dropbox.

There’s more to the news than just the connector, too. “We’re coming out with a plug-in and a set of prebuilt skills,” Zepf says.

All of which is a bit of a departure from Syncro’s previously stated intention to go all-in on Microsoft in cloud, security, and beyond. Hard to blame the company for diversifying its bets, though, given that while Microsoft remains a player in AI (Copilot seats grew an impressive 50% from 20 million to 30 million just in the last quarter) Anthropic is an even hotter one, with more revenue in the last 12 months than Starbucks, McDonald’s, and Taco Bell owner Yum! Brands.

Of course, betting on Microsoft, Anthropic, or anyone else in AI only makes sense for an MSP toolmaker in conjunction with a service desk automation roadmap, and Syncro’s other major announcement on Tuesday will be the launch of its first AI-powered ticket triage and dispatch solution. It will join a ticket summarization system introduced about six weeks ago that has improved time to action on technical issues by 10% on average, according to Syncro. Both features will be accompanied in September by a “guided ticket resolution” feature that the company projects will resolve 20% of tickets autonomously.

Competitors are claiming bigger achievements in that area, but aren’t leaning into chatbot integration as heavily as Syncro. We’re about to learn just how much MSPs value the ability to do their AI-accelerated managed services work through the same AI interface they use for everything else.

How the rich get richer

The answers to questions like that matter, according to George, and not just to vendors like Syncro. In the not too distant past, he says, security know-how was what most separated the fastest growing MSPs from the herd. In the increasingly imminent future, it will be AI.

“All of these technologies—these frontier models, the adoption of agent-based solutions, MCP—will be the fastest separator and the greatest separator this industry has ever seen,” George predicts.

If he’s right, it’ll be AI-forward giants like New Charter and Integris along with AI-native newcomers like Shield, Titan, and Treeline that benefit the most. Indeed, such companies are already using in-house expertise smaller peers don’t have and can’t afford to build sophisticated, AI-enabled operating environments, according to Slide’s Carlson, who notes that those are the MSPs in that 5-10% of Slide’s partner base using MCP to access data directly from their own heavily customized tools.

“Some of these folks don’t even necessarily use our interface,” Carlson says. “They’ve built their own entire monitoring dashboard across all of their products.”

The largest of them enjoy other advantages too, as I learned last week in a conversation with Ben Bernstein (pictured), manager of the cybersecurity advisors unit at Huntress, during the Black Hat conference in Las Vegas. There are three advisors in North America at present, each of whom provides direct, expert advice to about 10 of the continent’s biggest MSPs. Helping those companies be “a little bit quicker and a little bit more efficient,” Bernstein says, is the core of his team’s mission.