AI MRR FTW?

Hatz AI may have the first practical way for MSPs to make recurring revenue on artificial intelligence.

Just a few months ago, I shared an analyst’s prediction that it would be a good five to 10 years before managed service providers with SMB clients were making recurring revenue on artificial intelligence. Just last week, a startup named Hatz AI bet $2.5 million of seed funding that some MSPs could be earning MRR on AI this spring.

Or conceivably even this winter. Hatz AI plans to introduce the first two elements of what will eventually become a family of “AI-as-a-service” systems for MSPs some time in March (spring officially begins on the 19^th that month). Both are designed to empower MSPs currently using GenAI to accelerate scripting, automate ticket resolution, and otherwise streamline internal operations, if they’re using it at all, to collect one-time project fees and ongoing management fees from their clients as well in exchange for value-added, AI-based business tools.

The first tool, called AI Chat Assistant, enables MSPs to create role- and industry-specific AI companions their clients can use to shorten day-to-day workflows. The system will utilize OpenAI technology to start, with support for additional models to follow, but as far as end users are concerned it will all be powered by you.

“You brand it yourself. You put your logo on it, your colors, your fonts, all that,” explains Jimmy Hatzell (pictured), Hatz AI’s CEO and co-founder.

Partners needn’t school themselves in Large Language Models to use the product, he continues, and can collect high-margin fees for customization and deployment. “MSPs can manage it and make a margin off of each seat they sell as well,” Hatzell adds.

The company’s other initial solution, called AI App Builder, aims to put the power of generative AI in the hands of people who (unlike you, dear reader) haven’t spent a hunk of the last year mastering the finer points of prompt engineering.

MSPs will use the system’s low-code/no-code toolset to create apps. Their customers will use the apps to fill out perfectly ordinary online forms. Hatz AI’s software will automatically import data from the forms, use it produce a prompt behind the scenes, run the prompt past ChatGPT, and feed answers back to the customers in the app.

“They don’t even have to worry about what the prompt is,” Hatzell observes. “They’re just filling out the app, clicking yes or no, uploading a file.”

The revenue opportunity, as with chat assistants, is first in building the app and then in managing it. “You can think of it from the builder perspective as a UI for prompt engineering,” Hatzell says.

Both solutions will be backed by a multitenant management console, as will a third tool currently in pilot phase that will let businesses with overwhelmed customer service centers feed incoming calls to a voice-based, AI-powered answering service rather than a press 2, press 8, press 3 voicemail system.

Due toward the end of the year as well is an even more ambitious offering designed to let SMBs with privacy concerns or compliance mandates do what only enterprises can for the most part now—host custom LLMs of their own in a private cloud or public cloud instance. MSPs can charge per user per month for the service or on a consumption basis for the data users process.

All that’s interesting in its own right, but even more interesting when viewed as (potentially, at least) the first practical answer to a question MSPs spent much of 2023 asking: I know how to save money with AI. How do I make some too?

“Hatz AI really gives us an opportunity to quickly jump into the AI-as-a-service space,” says Michael Goldstein, president and CEO LAN Infotech, a Fort Lauderdale, Fla.-based MSP and Hatz AI early adopter. “Jimmy has a great concept to help us utilize AI, and then we provide potential AI-as-a-service solutions to our existing or new customer base.”

According to Hatzell, the long-term upside to this strategy is about as big as it gets in IT. “I think that this is a bigger opportunity than security,” he says. “It may take a couple of years to get there, but I really do.”

It’s easier to sell too, he adds. Security is an indispensable service that businesses hate paying for and hope they never need. AI is a competitive differentiator. “It’ll drive their costs down, it’ll drive their output up, where security is just a spend,” Hatzell says.

And the total addressable market is hard not to like. “It’s every business everywhere,” Hatzell says.

A perfectly good 2024 prediction spoiled

I admire people who not only make New Year’s predictions but score themselves on them later as well.

But why wait a whole year for that? A week ago, I forecast that accommodating AI data flows would drive money-making network infrastructure upgrades for SMB channel partners. A conversation this week with Bob Laliberte (pictured), principal analyst at analyst firm Dell’Oro Group, shot some holes in that forecast while adding nuance.

For one thing, Laliberte (pictured) notes, there are actually two topics at issue here: AI for networking and networking for AI. The latter, which is the one I had in mind, is about adding speed and capacity to networks moving AI data between cloud and edge all day. According to Laliberte, most organizations, and especially most small and midsize organizations, won’t need to worry about that because they won’t actually be moving very much data between cloud and edge.

“They don’t keep data at the edge for very long,” he says. “They’re processing it, getting the business insights they need in real time, and then typically sending that data back.” Hatz AI users with custom LLMs in a private cloud might be exceptions, but there won’t be a ton of them soon.

AI for networking (aka AIOps) on the other hand, while not a big deal for most SMBs this year either, could be a very big deal for MSPs soon. What if, instead of scrutinizing a network management interface or mastering a complicated command line interface, you could simply ask an AI networking bot in everyday English to tell you what’s slowing a network down and what to do about it? Some of the tools Laliberte has seen provide the right answer in seconds with about an 80% success rate.

“That’s not bad. It saves a hell of lot of time,” he says. “It means eight times out of 10 you’re not searching for what’s wrong. You know pretty much near real time that this is the issue and this is how we go fix it.”

A platform with automated self-healing functionality can even do the fixing for you, or better yet autonomously predict and address a network issue before it even happens. Laliberte has seen universities with self-healing AIOps solutions reduce tickets from 200 per semester to two.

“It’s able to find faults, fix faults, before the users even know what’s going on,” he says. And before technicians have to get involved as well.

“You can spend a lot more of your time focused on strategic initiatives,” Laliberte notes. “All of the troubleshooting, all of the configuration issues, and things like that are really going to be by the wayside.”

MSPs struggling to find qualified technicians will enjoy an additional benefit. “The ability to enable people with a lesser skill set to be able to effectively manage the environment, I think, is going to be really important moving forward,” Laliberte says.

If all that sounds like the future of networking to you, it does to network vendors too.  Extreme Networks and Aruba are two of several vendors with SMB-scaled products that offer AIOps. Aruba’s parent company, HPE, just spent $14 billion buying Juniper Networks and its Mist AI technology to get in the game too.

One last observation: networking for AI might not drive a lot of infrastructure upgrades, but security for networking for AI might. “Ultimately, when you’re building these models you want to be able to trust the data,” Laliberte observes.

We told you SASE rules the network.

It’s not just AI. Hyperautomation is hot too.

Saying so might discourage people from speaking with me in the future, but I’m starting to think that agreeing to an interview with Channelholic attracts severe weather.

The first evidence came last September, when CyberFOX found itself in the potential path of a hurricane after scheduling a meeting with me. The latest came this week when Aharon Chernin (pictured), the CEO of Tampa-based Rewst, booked time with me during what ended up being a tornado watch.

“Bear with me,” he said at the outset of our Zoom call while glancing nervously outside his home office window. “You could lose me at any moment.”

A disturbing prospect, no doubt, for Meritech Capital, which led the $31 million Series B funding round Rewst announced this week. That’s pretty good money for a company less than three years old under any circumstances, but especially good after an undeniably brutal year for venture capital.

It’s also a sign of the times. Artificial intelligence vendors like Hatz AI attracted fully a third of the $170.6 billion invested by U.S. venture capital firms last year, and hyperautomation vendors like Rewst clearly aren’t struggling to raise capital either.

Intriguingly, Rewst plans to spend the money not so much on artificial intelligence as on the human kind. Among other things, the company will be doubling its team of in-house consultants, who help MSPs put Rewst’s software to work removing manual effort from normally time-consuming technical processes. That hands-on assistance, according to Chernin, is a core part of Rewst’s value proposition. 

“We don’t leave the MSP to themselves to be successful,” he says.

Building out an integration library that currently includes 45 connections with solutions from Acronis, ConnectWise, Datto, Huntress, Microsoft, OpenAI, and SentinelOne, among others, is another priority for the new funds, along with completing the low-code application development tool Rewst introduced late last year, which is currently in alpha testing with 60 MSPs.

“Sometime early this year, you'll see us go from beta to production with it,” Chernin says.

All of that will come after a busy 2023 in which the vendor shipped 50 new pre-built automations, released an OpenAI-powered custom automation builder, and expanded Cluck U, its partner training and enablement platform.

According to Chernin, Rewst is spending a lot less time than it used to teaching MSPs the difference between the RMM scripting they’ve traditionally associated with automation and hyperautomation.

“We’re talking about linking three, four, five, half a dozen of your software-as-a-service products together to automate much larger things,” Chernin says. “It’s just a different type of automation.”

It’s also pulled in 700 partners pretty quickly. “The MSPs that want to be a leader in their space are looking for this,” Chernin says. “We don’t run ads and yet we’re growing so quickly that we did a Series B.”

Also worth noting

• On the eve of the biggest retail tech conference of the year, Microsoft has unveiled AI solutions for retail. So has Google.

• Robert Belgrave is Pax8’s new COO and Elizabeth McIlhany is its new chief product officer.

• Cloudli’s new CEO, meanwhile, is Jamie Minner, formerly of NetFortris and TPx, among others.

• Blancco has named Gagandeep Singh its VP of global marketing.

• Keeper Security users can now create and share granular permissions at the user level.

• UCaaS vendor Wildix has rolled out a new program that simplifies billing and payment while rewarding customers for paying early.

• Granite Telecommunications is now an official partner for Operator Connect for Microsoft Teams.

• The Alliance of Channel Women has five new board members, a new president, and a new vice president.