Exploring AI, Automation, and Team Dynamics in Cybersecurity for SMBs and MSPs with Gerwai Todd

[00:00:00] With as many breaches and security concerns as I report in this show, it should be obvious that cybersecurity is not just about technology, but also the human expertise needed to interpret and respond to complex threats.

[00:00:14] Huntress is focused on elevating SMBs and MSPs around the world.

[00:00:19] Huntress has a suite of fully-managed cybersecurity solutions powered by a 24x7 human-led SOC dedicated to continuous monitoring, expert investigation, and rapid response.

[00:00:31] And the proof is the execution.

[00:00:34] Huntress is the number one rated EDR for SMBs on G2.

[00:00:40] Want to know more about the platform? Visit huntress.com slash MSP Radio to learn more.

[00:00:47] Go ahead.

[00:00:48] Dave Sobel here with IT Nation Connect.

[00:00:51] Back again, frequent flyer, Gerwai Todd, who is CEO of both PIA and TimeZest.

[00:00:56] Welcome back to the show.

[00:00:57] So I want to, you've got this dual role for a little while, and I actually want to get a little bit of insight into how you approach that.

[00:01:05] Because it's unusual for a leader to lead two organizations at the same time.

[00:01:10] That's going to take a very particular kind of discipline and probably a very intentional approach.

[00:01:15] Give me a sense of how you approach that and how you're handling that.

[00:01:19] Dave I appreciate it, Dave.

[00:01:19] Great to see you again.

[00:01:20] Dave I think it starts with the team.

[00:01:24] I've got tremendous teammates on both sides.

[00:01:27] They know what they're responsible for.

[00:01:29] They take a lot of responsibility.

[00:01:30] We spend a lot of time in strategic planning.

[00:01:32] We do leadership meetings on a weekly basis.

[00:01:35] I'm a big EOS fan, and we've implemented that in both businesses.

[00:01:39] And that is a tremendous part of what makes it work.

[00:01:42] Dave In addition, it sure helps to have a team in Australia.

[00:01:45] So as the actor kind of gets along here in the US, my Australian team is coming online on the PIA side.

[00:01:51] And that makes it just a lot easier to kind of balance the workload, spread it out throughout the day.

[00:01:56] I just work longer hours.

[00:01:57] Dave That's an important detail because to recognize that you are actually splitting time then,

[00:02:00] and that you can work a longer day and cover two organizations.

[00:02:03] Dave I start on Sunday.

[00:02:04] Dave Okay.

[00:02:05] Dave I end on a Friday, do a little bit of work on Saturday, and it all works out.

[00:02:09] Dave That makes a lot of sense and good context for that.

[00:02:11] Dave So I actually want one of the things that we talked last year a lot about sort of state of the industry and thought.

[00:02:16] And I wanted to get a little bit more specific with our conversation this year.

[00:02:19] Dave Because particularly with what you're doing with PIA, feels like it's a space where you're thinking a lot about the combination of automation and AI.

[00:02:29] Dave And I want to get your sense as somebody who's kind of leading in that space.

[00:02:34] Dave I want to get first off a sense of where we are in practical adoption, that what's really happening versus what's hype on automation.

[00:02:42] Dave Let's set that ground discussion.

[00:02:43] Dave Give me a little bit of sense of what's happening out there practically.

[00:02:46] Dave Okay, sure.

[00:02:46] Dave Practically, numbers wise, we've had hundreds of customers using PIA on a day-to-day basis.

[00:02:53] Dave And so, you know, look, the way I think about the world is quite simple.

[00:02:56] Dave AI is just another tool for us software people.

[00:03:00] Dave So if you're in the software world, got tons of tools, tons of applications and languages we can use, AI just happens to be another tool in our toolbox.

[00:03:09] Dave Okay.

[00:03:10] Dave So it opens new doors in possibilities.

[00:03:13] Dave So effectively, we can solve new business challenges with this technology if done right.

[00:03:18] Dave And I really believe a big part of this process is ensuring that commercial viability of a product has reached the proper threshold.

[00:03:26] Dave So for example, AI hallucinates. You've got to knock down those hallucinations. You have to train it and really can't use these large language models.

[00:03:35] Dave These large language models are designed for these really broad use cases and purposes. They have value, but they don't necessarily apply to a particular industry specific use case.

[00:03:46] Dave Okay.

[00:03:46] Dave Like understanding an IT ticket. So if you think about a lot of what we do with models is we wipe the AI engine of its brains, right?

[00:03:53] Dave We remove all that internet of things in there, and then we program it with just the things we want to care about and know about. So as an example, as a ticket comes in on a help desk, we can train it to understand what that ticket's intent is, right? This is a passive reset. This is a new user. This is a PC issue that we may need to remediate through automation.

[00:04:13] Dave Okay. Yeah. That feels, I love that description of it because the training and the insight, what have you learned from teaching the models about what's really good in this space and what's also not good in this space?

[00:04:25] Dave Well, first and foremost, it's really hard. It's really hard work. You need super sharp people, people with experience in the space. And that is what I think is one of our core advantages as a team from a PIA perspective is the whole team from PIA has emerged out of the ITG in Australia.

[00:04:43] Dave It's much like the journey I ran on with Arnie, right?

[00:04:46] Dave Right.

[00:04:46] Dave ConnectWise, MSP, builds a PSA for themselves. And before you know it, all his friends and family wanted a copy effectively, right?

[00:04:53] Dave Right.

[00:04:54] Dave And so that's exactly what we've done. We've had the training base of an MSP to start with, and we've been able to laboratory a lot of that work. And that work started in 2019. It's been going on for quite a while.

[00:05:05] Dave And so today, what we've had is now we have a very mature model built over all those years, and we're able to also use other data inputs from our customer base and other sources to help improve that.

[00:05:18] Dave And what are the big wins? Because in a perfect world where I say, oh, it'll automatically handle tickets. We don't live in that world.

[00:05:29] Dave Sure. So give me a little bit of insight into the areas where the technology excels.

[00:05:33] Dave Yeah, it has to be boxed in. It has to have guardrails. But let me maybe go on a little bit of journey here. This may help because I think it's a big part of what we've been doing as an industry and it may add some focus here.

[00:05:46] Dave So I like to go back to the beginning. Like before when we used to have to roll trucks and when we were doing a lot of break-fix work. And you look at late 90s, remote control was the answer.

[00:05:57] Dave Sure.

[00:05:57] Dave So you could not have to roll a truck anymore. You could have a technician sitting at a desk and remote control.

[00:06:03] Dave And if you think about the ratio of the customers they could support, they could support about 100 customers per day.

[00:06:08] Dave Right.

[00:06:09] Dave 100 end users I can say.

[00:06:11] Dave Yeah.

[00:06:11] Dave But at that.

[00:06:12] Dave RMM came along. RMM plus remote control did great things, right? It added automation. You can kind of run things at the endpoint. You can get alerts in and be proactive about what's going on. And that allowed us to get to about 200 endpoints per day.

[00:06:26] Dave Okay.

[00:06:26] Dave But I'm then stopped, right? And I've actually been really tussling with like why did they stop, right? Why did they get gated at 200 and that became the ratio and then that's kind of all we did. And the reality is, at least my observation, is that the technology wasn't ready yet. The technology wasn't there to solve the next bastion effectively of automation possible.

[00:06:52] Dave Okay.

[00:06:53] Dave Okay. So I see that as the help desk. These service tickets that come in at a high rate, right?

[00:06:59] Dave Right. And are varying kinds of issues. We didn't have the technology at the time to do what we can do today. And number one is, we have to interrogate that ticket and we have to understand what its intent is.

[00:07:10] Dave Okay.

[00:07:11] Dave And with that, we can now begin to solve the equation of getting to 400 endpoints per tech. Because if we can scale at the service desk and we can make those tickets easier to work in a guided fashion, we can really solve a lot of challenges in doing that. So that's the mission at hand is how do you take those service tickets and get them resolved quicker with human observation, right? Effectively being that co-pilot for a technician.

[00:07:38] Dave Okay. You said something that's really interesting to me. It's understanding the intent of the ticket.

[00:07:43] Dave Yep.

[00:07:44] Dave Yep. Because what I want to square my understanding of that is, is one of the things that's interesting about the large language models is while they're really good at kind of processing and organizing information, they're not necessarily understanding intent. They don't necessarily understand what their report. They can reform information. We understand it and it is very useful, but the model doesn't understand intent. What do you mean by it has to understand intent?

[00:08:08] Dave So it starts with really good people on our side watching the model and how it behaves. In addition, getting inputs from our customer base as they use the application to see when we get it right and when we don't get it right and provide means for them to override our suggestion.

[00:08:29] Dave So let me just use it as an example. Imagine someone has a password reset, right?

[00:08:34] Dave Right.

[00:08:35] Dave That's pretty straightforward. There's a few words you have to read. You can pretty much assess the intent. What we do within the ticket is we prompt the technician with, we think this is a password reset. Maybe you agree, right?

[00:08:46] Dave Okay.

[00:08:46] Dave If they don't agree, if they think it's something else or some other job or action needs to be taken, we allow them to override us and then guide us through the proper next automation to run effectively.

[00:08:59] Dave And so our goal is to be a co-pilot with them. When they do guide us, we can see that. We can observe that they overrode us in this case. We can go back to the exact words that were in that ticket and then train the model to support that particular user.

[00:09:16] Dave Okay. Are you seeing patterns now in terms of the kinds of criteria that make this kind of technology really effective? Are there things that make certain kinds of requests really candidates for this?

[00:09:31] Dave Yes. Look, if we could have everything come in through a form, we'd be in a beautiful space because it's a controlled environment. We're asking for key inputs of what we need to understand the issue and then we can take action upon that.

[00:09:44] Dave Okay. Dave Free-form email is a dangerous game and we all know that. It's been a dangerous game for a long time. But what we've been able to do and what these models are capable of doing today is really getting quite accurate at that. Our confidence level keeps growing every day about what we know about a particular ticket. And again, the key here is it's like a Tesla, right? We have a human driver in there overseeing what things we suggest.

[00:10:11] Dave Gotcha. That makes a ton of sense. So speaking broadly, and I guess I'll also say like knowing I'm not asking you to make any product promises.

[00:10:19] Dave Okay.

[00:10:19] Dave But give me a bit of a sense of the where you see this capability going. Like what would be like an in-state model from a vision perspective?

[00:10:29] Dave Yeah, good question. Look, we have a lot of automations in our businesses and there's a lot of things that can potentially trigger those automations.

[00:10:39] Dave The more intelligent we can be about the triggering of those events and as our confidence grows about having it right, there's going to be a point where there's going to be less and less touching of tickets and way more direct automation happening in the process.

[00:10:58] Dave And it'll happen in stages, you know, as we mentioned, a passive reset is pretty obvious, right?

[00:11:04] Dave Sure.

[00:11:04] Dave In fact, a passive reset coming through a form, quite obvious, right? What needs to be done there? A new user ad coming through a form.

[00:11:10] Dave We can execute everything that needs to happen upon submission of those kinds of forms. And we built the framework to be able to support that.

[00:11:18] Dave The hard work really is being able to configure to all the different various possible environmental things that are going on in the infrastructure, right? So you could have Microsoft 365 in the cloud plus an on-prem server that still needs to be synced.

[00:11:33] Dave Right.

[00:11:34] Dave We've got all these different various hybrid configurations. And fortunately at this stage, because of the time put in, we've configured and we're capable of supporting all those different.

[00:11:43] Dave It's interesting you talk about that because it feels like one of the promises of AI, of course, is the idea of you can just throw it out and structure data and it will just magically do stuff, right? And your face gives it away, right? Your face gives it away. While it's nice and intention, what we're actually finding with a lot of the key implementations that are working is when there's some level of organization to the data, things go better.

[00:12:04] Dave When we have a level of data governance. Give me a little bit of a sense of what you've learned about what solution providers themselves should be thinking about in terms of their own data that make them more ready for this kind of capability. Like what do they need to be doing to make their own data houses look right?

[00:12:24] David You know, what's interesting is one of the very first things we can do a really, really good job with this technology is setting type, subtype, and item on a ticket, right?

[00:12:32] Dave So if you think about that, we're at the stage where we can establish that. But if someone hasn't configured their type, subtype, and item to some standards like ITIL or otherwise, we've got quite a bit of work to do to just bring up the process standard to then be able to properly assign the categories to these articulately.

[00:12:51] Dave Got it.

[00:12:52] Dave So that's one step.

[00:12:54] Dave That is super helpful and also gives everyone a good starting point to think about. I really appreciate you joining me. You always can make such good insights into what's happening in the field. This has been great to catch.

[00:13:02] Dave Good job. Thanks for joining me.

[00:13:03] Dave I love it. Great.

[00:13:05] Dave Today's episode is supported by CoreView. Your customers need your Microsoft 365 expertise and CoreView has the only M365 management platform designed for MSPs. Manage hundreds of tenants, automate manual tasks, and monitor compliance, all while intelligently comparing to the baseline. With a no-code control approach, CoreView revolutionizes your Microsoft 365 administration.

[00:13:30] Dave This powerful platform enables automatic reporting and remediation, ensuring optimal performance and security. The best part? You achieve this high level of service without the need for a large workforce, allowing you to focus on growing your business through efficiency. Want to know more? Visit coreview.com slash MSP and find out more.

[00:13:57] Dave Sobel

[00:13:59] Dave Sobel Under ethics guidelines posted at businessof.tech

[00:14:03] Dave If you like the content, please make sure to hit that like button and follow or subscribe. It's free and easy and the best way to support the show and help us grow.

[00:14:13] Dave You can also check out our Patreon where you can join the Business of Tech community at patreon.com slash MSP Radio or buy our Why Do We Care merch at businessof.tech.

[00:14:26] Dave Finally, if you're interested in advertising on this show, visit MSP Radio or buy our Why Do We Care merch at businessof.com slash engage.

[00:14:33] Dave Once again, thanks for listening to me, and I will talk to you again on our next episode of the Business of Tech.

[00:14:42] Dave Part of the MSP Radio Network.

[00:14:46] Dave