Atera’s AI is Not Getting Older. It’s Getting Better.
The vendor’s agentic automation tool hints at what’s ahead for younger peers. Plus: Why Squash put governance ahead of AI, why Todyl values GRC proof, and who made the 2026 ConnectWise PitchIT list.
Channelholic regulars are thoroughly familiar with how fascinated I am by AI productivity solutions for MSPs. I write about them all the time. Some are made by startups fighting for market share and brand recognition. Others are from the biggest names in the industry. Some (like Squash, of which more in a moment) have been around several months. Others (like Pia and Thread) have been around several years.
When it comes specifically to agentic AI productivity systems for MSPs, though, the very first you’ve read about here through the years is Atera Autopilot, or Robin as it’s been named since March, which first began stirring to life in the company’s R&D lab five months before ChatGPT’s debut and first appeared on my radar some 13 months after that milestone.
That makes it, by the standards of its product category at least, old, a trait it happens to share with yours truly. And just as my increasingly gray hair is a harbinger of what younger readers have ahead of them, so too does Robin offer hints about what lies ahead for younger autonomous MSP solutions. I count at least five clues about what the future holds for such systems as they mature, in fact, based on a recent conversation with Atera CEO Gil Pekelman (pictured).
1. They’ll get smarter. With AI as with people, age breeds experience, experience breeds context, and context breeds wisdom. Robin is no exception.
“Robin processes about a million events every month that increase its learning, its capabilities, its accuracy, and its speed,” Pekelman says. “It gets much smarter all the time.”
It’s smart enough already, in fact, that its ability to close tickets without human help has progressed beyond the numerous Tier 1 issues it initially addressed, according to Pekelman, “deeply into Tier 2.” Indeed, Robin’s know-how extends so deep into Tier 2 at this point that Atera now guarantees partners they’ll be autonomously resolving 50% of all tickets within 90 days of onboarding.
2. They’ll get faster. Robin doesn’t just know more than it used to. It moves faster too, according to Pekelman, who says mean time to repair end user problems now averages 120 seconds.
3. They’ll deliver steadily increasing ROI. In addition to getting smarter and speedier with the passage of time, Robin’s growing more scalable too. Atera has partners supporting 1,000 users per technician at present, and the number continues to climb. That’s a boon to MSPs pursuing richer margins.
“You can grow without adding more people,” Pekelman observes.
4. They’ll transform what technicians do for a living. Most Robin users, Atera says, reassign one-time help desk technicians to new, more valuable roles. Sometimes it’s high-margin consulting and project work. Other times it’s an entirely new function brought into being by agentic AI, which Microsoft calls “agent boss” and Pekelman calls “agent manager.”
“It’s a new job, a new profession that’s developing,” he says. “We’re seeing it more and more.”
Robin, he continues, isn’t so much an agent as a swarm of agents with different skills and responsibilities. “You have a lot of them, they do all kinds of things, and you need to teach them,” Pekelman notes, not just about what they must and must never do, but about the right and wrong ways to handle permitted work as well. Agent managers are responsible for that work.
5. They’ll grow easier to use. Robin’s not hard to use now, according to Atera, but will require even less expertise when a major new update rolls out soon.
“We’re working on the next generation of IT management platforms, the agentic RMM and PSA,” is all Pekelman will say about it for now. Publicly, anyway. He dropped some large and interesting hints about what’s coming off the record, and I’ll share the whole story here just as soon as the product ships.
What hasn’t changed
Usage-based LLM pricing schemes, as I’ve written here before, make setting budgets and controlling margins difficult, and one could easily imagine vendors passing that headache along to their partners. Atera has decided not to. Robin, which currently costs $20 a month per authorized user, won’t be switching to consumption-based rates any time soon, according to Pekelman.
“We decided to take the path of giving our customers visibility and confidence in how much it will cost them as opposed to connecting it to tokens,” he says, noting that Atera can afford the risk of mixing variable upstream spending with flat-rate downstream pricing thanks in part to Robin’s growing age and wisdom.
“We have enough information now to make sure it’s a calculated risk,” Pekelman says.
It’s governance first, AI second at Squash
Atera’s guaranteed 50% automated ticket resolution rate within 90 days of implementation is essentially just a starting point. Robin can potentially close as many as 92% of tickets eventually if you grant it permission to address all the many ticket types it’s capable of addressing.
“There are customers that do that and there are others that limit it, and they’re at 75% or 60%,” Pekelman says.
Hard to know without guessing why users who limit Robin’s actions are doing so, but uneasiness about the many things that can go wrong when AI acts alone probably has something to do with it, suggesting that there’s a direct relationship between how much MSPs trust their AI automation tool and how much productivity benefit they get from it.
Which brings us to Squash, a very young service desk automation startup founded by veterans of Silicon Valley high-flyers Rippling and Harvey that’s betting big AI can deliver a lot more for MSPs than it is today provided those MSPs are comfortable allowing it to.
“We are at a point with the technology where the tech is not the limitation any longer,” says CEO and co-founder Sandeep Uppaluri (pictured). “The limitations that we see are basically MSPs’ willingness to lean in, adopt, and make the most of the tech.”
The product Squash built in response to that analysis stands apart, the company believes, for two reasons, starting with the breadth and depth of its automation abilities.
“A lot of the other players out there in the market today are mostly focused on ticket intake, maybe triage and dispatch of tickets,” Uppaluri says. “We integrate very broadly into your entire tech stack, and we will take actions on your behalf to help you resolve those tickets.”
More importantly, he continues, AI governance is a core design pillar rather than afterthought at Squash.
“We built out the governance controls before we built out any of the AI stuff, basically because we think it’s super important,” Uppaluri says. “The better your governance controls, the more it allows you to unlock the power of the models, the further you can push them, and the more you can get out of them.”
Squash uses a containerized environment within the tool to enforce governance. Incoming tickets go straight to that isolated space, where the system automatically applies the user’s guardrails. The system helps users slowly define those policies over time on a ticket type by ticket type basis.
“A lot of our product is designed around how we can enable a gradual rollout,” Uppaluri says. “It’s that trust building that is the crux of getting value out of AI. It’s not tech.”
Squash users come in every variety, the company says, but tilt a bit toward the kind of MSP running OpenClaw in the lab because they like living on the leading edge.
“Very quickly they start to realize that you need more governance controls,” Uppaluri says.
That Squash takes such controls seriously isn’t the only thing distinctive about it, however, he adds. The company’s unusually flexible pricing scheme sets it apart as well, not just from other AI startups but from nearly everyone else as well. Want per endpoint pricing? Squash is happy to provide it. Prefer something outcome-based? Squash has you covered there too.
“We try to flex the model to make sure that it’s working for you,” Uppaluri says. “The goal is to make it feel as bespoke as possible.” And to ensure that the fees partners pay and productivity gains they collect net out to about a 20% increase in service margins, he adds.
Consistently. Like Atera, Squash has opted not to saddle partners with the challenges of earning predictable profits while paying unpredictable token-based rates.
“MSPs shouldn’t be worrying about what model to use, how much it costs, how the price has changed, and how to orchestrate these different things,” Upplauri says. “That’s the responsibility of a vendor.”
Don’t shout AI
Squash is one of many vendors touting AI-based managed services functionality these days. Uppaluri is the first to admit that sorting through them all is hard work for MSPs.
“Every vendor claims to have AI solutions,” he says. “There are just a lot of different things coming at them in different ways.”
Sean Lardo feels their pain. Every spring for the past several years, he’s led a committee at ConnectWise tasked with deciding who gets into the company’s annual PitchIT program, a 16-week startup incubator that hands out $150,000 worth of prizes to the three best graduates.
“It was very challenging this year,” says Lardo (pictured), ConnectWise’s vice president of IT Nation Communities. And not just because the applicant pool was unusually deep (over 65 companies competing for what turned out to be 25 slots).
“These companies are claiming AI in everything,” Lardo says, adding that it didn’t impress the judges. “You shouting AI doesn’t mean you have anything. It’s about what you’re doing with it. What’s the actual use case?”
The use cases Lardo and his fellow evaluators prioritized reflect their read on what MSPs most want and/or need right now. “If you look across the board for this class, everybody represents some segment or fraction of the business that’s a necessity or will be more of a necessity moving forward,” Lardo says.
Some of those necessities relate directly to AI itself. Class-members Synthreo and Aportio, for example, make AI-powered service desk automation software while Lemhi, which you met here recently, helps MSPs sell customer-facing AI services. Kipling Secure’s solution, meanwhile, aims to help users mitigate the risks those AI services can pose.
“AI cybersecurity is big because AI hacking is a real thing now,” Lardo says. “It’s happening faster, more often, and it’s really hard.”
Other contestants, including Handover, TechAlign, and MSP Report Card, help with more traditional tasks like conducting quarterly business reviews, something too few MSPs do at all, according to Lardo, and fewer still do properly.
“They don’t use them for what they should be, which is an actual growth tool, a sales enablement tool,” he says.
Lardo advises MSPs struggling to evaluate the many young vendors vying for their business to ask the same questions he and his team did when vetting would-be PitchIT contestants. Does the leadership team have experience in managed services? What do people say about the company online? How long has it been around?
“You have to look at their credibility,” Lardo says. “We don’t expect ten years of business, but we do expect to see a couple of years of them doing things.”
ConnectWise will name this year’s three finalists in September. Two months later, a panel of judges from outside the company will crown the winner at the vendor’s IT Nation Connect event in Orlando. Which contender grabs their eye most will offer an interesting glimpse at what a winning recipe for success with MSPs looks like in the age of AI.
Speaking of AI-era startups…
The founder and CEO of one, Cyft, joined us on the most recent episode of MSP Chat, the podcast I co-host, for a thoughtful look at the present and future of AI-powered productivity for MSPs. Give it a watch here and check out all of our other interesting guests here.
Todyl offers proof, not promises
AI and security solutions, predictably, figure prominently on this year’s lineup of PitchIT contestants. So too, somewhat less predictably, does QBR software. Another issue ConnectWise apparently thinks MSPs want and need help with right now, though, is compliance, which shows up in products from three of the 25 vendors on this year’s PitchIT list. Not surprising, according to Lardo, given how many rules, requirements, and regulations there are to comply with and the potential consequences for MSPs and end users who slip up.
“They could lose their whole business,” he says.
Todyl is aware of that danger too, as well as the burdens associated with avoiding it. Cyberinsurers and regulators want more than your word on compliance, observes Ken Patterson (pictured), the security vendor’s VP of community.
“They’re asking for proof,” he says. “MSPs, quite frankly, are not ready for this.”
Todyl can help, Patterson says, both through its platform, which imports reporting data from the vendor’s SIEM, MDR, and other products directly into its GRC solution, and through alliances with the partners featured on its recently introduced Assurance Marketplace.
There are four such organizations at present, each of which offers something cyberinsurers and other compliance enforcers value. Membership group GTIA offers third-party validation of cyber readiness through its Cybersecurity Trustmark program, certification and cyber warranty provider SPECTRA provides third-party validation of security best practices, insurer Ledgebrook offers better rates and coverage based on validated security evidence, and Optimize Cyber delivers pen testing and incident response services many privacy and safety regulations require. Todyl may add more companies, according to Patterson, but no more than a carefully curated few.
“We’re not looking to be a distributor,” he says.
The Assurance Marketplace is part of a broader MSP push at Todyl, which sells exclusively through the channel. Other examples include the virtual CISO it’s appointed to help partners advise clients and new courses it offers on delivering AI security services.
“My whole thing is about educating and enablement,” Patterson says.
That and community, which is closely tied to enablement in Patterson’s mind. Vendors often call the partners who sell and support their products their community.
“That’s not a community,” Patterson says. “We have to give back. There has to be a back and forth relationship between us and them.”
Preferably in person whenever possible. AI is sure to make vendors like Todyl more effective in the years ahead, says Patterson, who spoke with me at Rewst’s FLOW event in Nashville last month.
“It’ll never replace this though,” he said, pointing back and forth at us, “the direct face-to-face conversation.”
Over on The Business of Tech
Host Dave Sobel was joined live by yours truly last week to discuss recent news from Pax8, ConnectWise, and beyond. An interesting conversation, at least from my end. See if you agree.
Also worth noting
Businesses added 47,000 tech workers last month, according to CompTIA, down from 69,000 in May. On the other hand, tech unemployment is down, too, from 3.1% to 2.9%.
New data from AvePoint suggests that AI agent adoption is accelerating faster than governance. Not surprising, perhaps, given that MSPs aren’t jumping to help.
CyberFOX has equipped itself to offer cloud-native SASE and zero trust network access by acquiring Timus Networks.
The new Managed Identity Security Posture Management solution for Microsoft 365 environments from Huntress is now generally available.
New research from Cynomi reveals the top AI questions MSPs are asking in 2026. Number one on the list? How do we say no to client AI requests without losing the account?
MSP360 says it will not increase pricing on its managed products for eligible MSP partners. Ever.
DNSFilter’s new MSP Partner Program offers enhanced benefits, resources, and incentives intended to help members grow their security businesses.
SnapLogic’s new MCP Builder is designed to turn existing integrations, APIs, and business processes into governed MCP tools in minutes.
BlueVoyant has introduced a dedicated AI agent security service for Microsoft environments.
Exabeam has expanded its behavioral intelligence capabilities to detect, investigate, and reduce risk from AI agents, autonomous workflows, and human-to-agent activity.
New capabilities added by Infoblox to its Universal Asset Insights solution are designed to improve visibility for AgenticOps initiatives and IT automation.
MyCloud Managed Security, from Reinvent Telecom, is designed to help partners expand into managed cybersecurity services.
Intruder’s new free plan provides permanent access to core exposure management capabilities.
Impartner has added native CRM synchronization for HubSpot to its PRM platform.
Mika Yamamoto is the new chief marketing and customer AI officer at Veeam.