‘Embrace the tools,’ says ESET’s Ryan Grant. ‘But at the same token, have the technology that’s actually treating the AI tools and the agents almost like an employee.’
While new artificial intelligence technology offers efficiencies in security such as cutting down on alert fatigue, security vendor ESET still emphasizes humans in the loop, Ryan Grant, ESET country manager for U.S. and Canada, told CRN in an interview.
“We have 24-by-seven monitoring that’s watched over by a human,” Grant said. “We’re not taking our hands off the wheel completely. … Our recommendations for how people should be thinking about the tools is, yes, embrace the tools. But at the same token, have the technology that’s actually treating the AI tools and the agents almost like an employee.”
ESET is “not trying to slow down the advancements of using the agents,” Grant said. “However, you should be testing those agents to understand what skills it is [using].”
The Bratislava, Slovakia-based cybersecurity vendor positions its new Protect portfolio of products and services as a way to help bring its enterprise-grade security offerings, including that 24-by-seven monitoring, downstream to small and midsize businesses, helping SMB-focused solution providers protect clients against the higher volume of cyber attacks and new attack surfaces in the AI era.
[RELATED: Delivering Reliable Connectivity And Cybersecurity On The High Seas: Inside MSP Marlink’s Approach]
ESET has about 10,000 partners worldwide, according to CRN’s 2026 Channel Chiefs.
Brad Kostner, executive vice president of sales at Phoenix-based ESET solution provider Adrytech, told CRN in an interview that educating customers on new AI threats like copying voices for phishing and stealing credentials—a practice called vishing—has been in high demand.
Helping customers navigate the latest security technologies and best practices also remains a strong part of the business, especially as the volume of phishing attempts grows in the AI era, Kostner said.
“We can put as much MDR, high-level security statuses, different rules in place, but if you have end users that are still clicking on things that they’re not supposed to, all [the tools] in the world are not going to help you,” he said.
Here is more of what Grant had to say about ESET’s new Protect portfolio.
What’s the message to ESET partners looking to leverage the new Protect portfolio?
When you look at the market overall, everybody is extremely overwhelmed with what’s going on with security, particularly now with AI. … Earlier this year, we actually did an SMB survey.
We polled a number of SMB organizations, and it clearly came back that people are very overwhelmed. They can’t keep up with cybersecurity. … We’re really trying to offer effortless cybersecurity for everybody in this AI era.
And to be honest, it’s really to drive more simplicity for a very complex security market right now. And that’s effectively what we’ve done.
We’ve basically looked at all the requirements that are needed out there. And we’ve integrated everything from all of our AI to all of our 24-by-7 monitoring to vulnerability and patch management to cyber warranty.
We have bundled it all into a very simple offering. And it’s not stitched together with separate consoles. It’s really one system. It’s all natively built in.
And it really is to allow for people to effectively just have cybersecurity in a box. Almost like cybersecurity as a service. It has all the features there. And really just truly simplify it for the partners and also simplify it for the clients.
Is AI at the product level a new investment for ESET?
We’ve been on the AI side of things since 1997. So we’ve had advanced heuristics and large language models built into the platform for quite some time.
We’ve built upon it. … Back in 2023, we actually launched an AI Advisor … for people that don’t know what’s happening within the SOC [Security Operations Center]. … They can ask it, ‘What’s happening to my environment?’
And it will clearly articulate what’s going on, what to do and how to remediate. … The big features that are coming from an AI perspective are really three areas.
One is AI browser and conversation security. So think of users going out, using OpenAI or Claude or these sorts of things.
And when you think about it from a business standpoint, that can be quite concerning if you’re uploading information that’s company proprietary information or it’s sensitive.
We basically are protecting that entire environment, including URLs and scripts and those sorts of things, to make sure things aren’t uploaded that shouldn’t be uploaded. … The other part of it as well is AI agent security, which I think most people are probably most concerned about. … The AI agent, we look at the intention of what it’s supposed to do. Help define the guardrails.
And then not only do we check what it’s doing, but then we monitor the behavior of the agent. And that’s where it gets tricky. You can look at an agent. And you can look what it’s supposed to do. And then you could technically deploy it into production. And that’s great.
But if you’re not monitoring it like another employee, that’s where it becomes problematic. So our recommendation for how people should be thinking about the tools is, yes, embrace the tools. But at the same token, have the technology that’s actually treating the AI tools and the agents almost like an employee.
What access of data do they have? What’s the governance of it? And those sorts of things. Treat it as identity, almost to that degree. And making sure that it’s actually doing what it’s supposed to be doing while also staying protected and having the guardrails.
What should ESET partners know about how you’ve packaged these tools?
We only do cybersecurity. That’s all we do. And we’ve got to make this very simple.
We didn’t acquire any AI technology. It’s all native. And so it’s all natively built into the platform. So for us, it wasn’t like, ‘Look, we’re going to try to monetize this.’
We’re basically going to just build the features into the product. And we give people some different tiers to choose from. But we’re just going to include it in.
So when you, as a user, are going to prompt our AI, you’re not going to get hit with tokenization. … We’re just going to include it all in [our offering]. … It is coming from a place of trying to make it easier.
The key thing for us is, when we did the survey and we looked at what was happening in the SMB market—these SMBs, they have probably the biggest challenge, more so than enterprises and midmarket because they have the budgets to do it.
So our whole idea is we need to have a comprehensive offer that the SMBs can use at an affordable price but also have the best technology that’s there.
Are you seeing security vendors look to tokenization for increasing revenue?
It actually has started. … We’ve heard from clients. … They are a bit surprised on the cost of AI from prompting and the tokenization, the impact that’s having.
It’s definitely a factor. We did not want to go that route. But I also think part of that, too, is the fact that we didn’t acquire the technology.
It’s all natively built. We don’t have things like payback period and these sorts of things that you’re going through if you’re acquiring it and then you’re trying to stitch it together to integrate the tools together.
We just don’t have that cost structure. And given we’ve been in AI since ’97, for us we had a really strong foundation to begin with, and we just basically built upon it.
Are the reports on AI agents going rogue impacting customer appetite for the technology?
A lot of people were downloading … AI agents [when they first came out] and deploying them in production. And they already had malicious files in there. … We’re not trying to slow down the advancements of using the agents.
However, you should be testing those agents to understand what skills [they are] using. We actually have a free AI skill checker on our website that we offer to people.
In a very short period of time, it was like two months, we scanned like 900,000 skills—25,000 of them were suspicious. And 3,000 were malicious. … We’re actually looking at multilayer defense.
We’re unpacking what the agent’s doing. And then we’re actually observing what it’s doing in a sandbox environment. … If we feel like it’s clean, then we can push it into production.
And then once it’s pushed into production, we can actually see and make sure it’s actually staying within the guardrails that were defined, that we saw in the testing area.
You have got to have the security to monitor the behavior, but you really actually have got to take a look at that AI agent and the skill that it’s deploying and making sure it’s doing what you intended it to do.
Where it becomes important for the MSPs is making sure you have data governance and cyber protection on these agents and really treating it like an employee.
That’s where the opportunity resides for the channel. It’s a huge opportunity for them.
Was it surprising to see this early embrace of AI agents?
Everybody’s using these for efficiencies, which is great. … The challenge lies in the fact that it’s shadow AI where it’s unmanaged and it’s creeping. … If you have disparate tools out there, our systems can monitor what’s happening there.
But in terms of controlling it, this falls back on fundamentals. Have an AI policy, pick a browser to use and a motion to use. And then train the organization on that. … Our platform is really designed to effectively just look at all the conversations that are happening.
So it’s really focused on between the user and the large language model. It scans URLs. It scans scripts. It flags any kind of sensitive data in the prompts. And then any kind of potential uploads, it’s also scanning for that as well. … That’s exactly what we call ‘AI conversation’ or ‘AI browser protection’—that’s what it’s doing.
But there are ways to really shrink the attack surface if you have set policies and set tools in place with the organization. That’s where the MSPs really come into play.
How is AI impacting attack surfaces?
The attacks are getting faster. The deceptions are getting better. But actually the attack surface is getting bigger because AI is getting into spots that you wouldn’t have thought before.
We have actually seen in our research AI decoys. … AI is making its way into different areas, and now the attackers are using it as well as the defenders. So people are getting more creative with it.
Do you see the current growth opportunities for MSPs extend beyond AI?
People are worried about AI, but some of the most basic fundamentals that you would see in like a NIST [National Institute of Standards and Technology] framework or CIS [Center for Internet Security] framework are not being addressed in these SMBs.
It is a little scary. But at the same token, from an MSP standpoint, huge opportunity. We get a lot of questions with regard to how MSPs should handle this. … If you’re in the space, really have your house in order.
So the MSP should have their house in order around governance and mapping. So things like have an allowed-AI registry. What tools are you going to use? Have data boundaries in terms of where your customers’ data is going to sit.
And then have an audit of your RMM [remote monitoring and management] and your PSA [professional services automation] tools. … We’re a big fan of, ‘Eat your own dog food.’ And so we’re encouraging our partners to leverage our offering, what we have for AI, into their infrastructure. … You don’t really need to build an AI practice.
Once you’ve got the fundamentals in place, we’re telling MSPs, ‘Extend the cyber. Extend the governance. Extend the access control.’ … And then help the organizations around AI readiness assessments. Help them draft the policies. And then you can’t overtrain your teams enough.
And there’s a way to monetize that for the MSPs. So there’s a huge opportunity out there for them.
At the end of the day, on the product side, we’re trying to keep this very simple, even for the MSPs. Because the MSPs have the same headaches as the end customers do.
This product is designed from the channel. We’ve been a channel partner since day one. We took the feedback from them. We applied it. But it’s intentional in the fact that they can use it as a partner or MSP, and they can extend that same technology down to their client.
How is AI impacting your technology?
The platform, it’s 24-by-7, AI-led human monitoring. Most hacks are happening when we’re asleep—2 a.m., 3 a.m.
We have 24-by-7 monitoring that’s watched over by a human. We’re not taking our hands off the wheel completely.
We also include vulnerability and patch management. One of the biggest reasons that people get hacked is they miss an asset. They don’t do a patch update or system update. So we’ve included that in there as well.
Multifactor authentication is included. All the endpoints are protected. If anybody’s using a Microsoft environment, 365, or Google Workspace, that is protected as well.
And then we wrap it in premium support. And we wrap a cyber warranty on top of it.
It is all-inclusive. If they need to go get an insurance policy, they can just lead with our offering. It’s got a warranty tied to it, so it’s everything that most companies are going to need.
How big of a deal would you say the new packaging is?
What we’re doing is a game-changer for the industry because we’ve really simplified it.
I confidently feel like we’re leading the way in terms of really getting people to have an end-to-end, comprehensive security posture and not breaking the budget.
But at the same token, making it easy for them to actually get an insurance policy if they need to get an insurance policy. … Once they have a problem, they’re like, ‘Oh, I probably need to extend my cyber protection a little bit further.’
So we’re trying to basically educate the market, get ahead of it. SMBs are very vulnerable out there. They’re still the backbone of the American economy. And so we’re really trying to help with that.
How does ESET view giving security work to AI versus keeping humans centered? Where can AI automate away work without much risk?
The human interaction is critical. I always tell people … ‘Do you want to take your hands off the wheel and just let the AI do whatever? Or are you completely on the other end of the spectrum?’
Our position here is lead with AI, leverage AI and the efficiencies you gain from it. But we need to have human oversight. And that’s what we’ve done with our SOC. That’s what we’ve done with our research and everything else. …
So that’s always been our approach. … The beauty of this is that if you’re running a SOC or have a SOC is the efficiency you gain.
You hear it—alert fatigue. ‘I’m overwhelmed with all these things.’ So that’s where AI really drives efficiencies.
And you become not reactive. Now you actually become proactive as a SOC operator. … AI has enabled that [enterprise-level tier-one SOC environment] to come downstream. … One of the big things with our launch here is that you don’t need an enterprise budget to have an enterprise-grade protection environment. … If you’re a 50-seat SMB, if you are 500 seats, or 5,000 seats, you’re getting enterprise-grade security.
