The Future of IT & AI with Oshri Moyal, CTO & Co-Founder at Atera

(Intro Narrator at 00:00:00) Today, we're talking to Oshri Moyal, CTO and co-founder at Atera, about how AI is transforming the world of IT. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:16) I read your bio, dude. It's crazy. You snowboard. You have five AI patents. You're in Israel. You're a father of three. That's unbelievable, man.

(Oshri Moyal at 00:00:26) You know, all the fun stuff that you mentioned, being a father, it's a full-time job.

(Joel Beasley at 00:00:32) How old's your oldest?

(Oshri Moyal at 00:00:33) 10.

(Joel Beasley at 00:00:34) Okay. I've got three and my oldest is seven. So I'm right there with you.

(Oshri Moyal at 00:00:40) Amazing.

(Joel Beasley at 00:00:40) So tell me a little bit about your patents. You've got five of them. What are they around?

(Oshri Moyal at 00:00:45) Actually, everything is around exactly what Atera does, the AI-powered IT. And it actually started long time ago. There were all these buzzwords of big data and machine learning and everything, and we noticed that we have over a million tickets every month. So the first time, we felt like there is something big in the data, and we have an opportunity because we were the first and only all-in-one solution for IT technicians. We're combining the ticketing and the help desk with remote monitoring. We, from one hand, have all the tickets and the communication between the clients and the IT technicians. And on the other end, we have all the remote sessions, how IT people are solving problems and what they are dealing with every day. And I think the initial idea was, what if I can just film everything, just record their day-to-day, see what they are dealing with, and try to automate it. You know, take those repetitive, mundane tasks that they are doing all day, which is also something that we realized by analyzing the data. We noticed that they are fixing the same issues over and over again every day.

(Oshri Moyal at 00:02:14) So the first patent was, you know, go and record everything, do machine learning algorithms over it, and automatically solve it the next time it happened.

(Joel Beasley at 00:02:26) That is awesome. And now your company is doing that for all types of companies. Is that correct?

(Oshri Moyal at 00:02:31) Exactly.

(Joel Beasley at 00:02:32) Yeah. So is your customer the IT person? A lot of the CTOs, technology leaders listen to our show. Should they be interested in it, or is it another buyer within the company?

(Oshri Moyal at 00:02:45) Absolutely. CTOs, IT technicians, system administrators, all the tech guys. But we noticed lately, especially when we're talking about AI and the efficiency it brings to the table, that also CIOs and even CEOs are relevant, because, you know, everybody wants an efficient company with efficient processes and now get more from their employee time.

(Joel Beasley at 00:03:17) So if I'm a company and I have an internal IT help desk team, and they're responding to the same things over and over and over, I might have 50 of these people. Right? The AI can—how does it work when they adopt the technology? Do you train it on their existing information? How does the implementation actually go when you get a new customer?

(Oshri Moyal at 00:03:41) So first of all, we have two kinds of solutions in the system. We have, we call it Copilot and Autopilot, which are completely two different apart. We'll start from the Copilot. Copilot is just an AI assistant that's integrated into our platform, and it helps the IT technician doing the task, the day-to-day tasks, the complicated work. If he needs to write script, the AI will do it for him. If he needs to configure some very complicated SNMP device, you just type exactly what you want and the AI will do it. When he's logging into a machine and typing command lines and he needs to do something complicated, the AI will be there. Every time he solves a ticket, every time he's seeing an alert, the AI will be there to suggest what is the best solution, and will do also some analysis. So the AI can connect to the remote machines, connect to Office 365, Azure AD, gather all the information, and give the best solution. And the solution won't be just a verbal solution or text. It would be a button, or even the AI will do the action for him.

(Oshri Moyal at 00:05:01) So we just connected everything into our core technology and let the AI connect using our tools and do the actual work. So this part is only the Copilot, which is the AI assistant for IT technicians. On the other hand, we have Autopilot, which is actually an AI IT technician that gives first-tier support to the end users. So every employee in the company will have a desktop widget in his machine, and every time he has some issue or request or just even a question, he just interacts with this chatbot. But this chatbot is also connected to Atera platform. So if you complain about your computer is slow or you can't print, or those are actually the things that happen every day to almost every one of us, the AI will connect to your system, check, and try to find the root cause of the issue. We'll run some diagnostics on your machine and then offer a solution, and we'll perform it for you as the user. And you mentioned if we are connecting to the data and we are learning from it, we are also connected to the company's knowledge base. So every time the technician solves a ticket or has something very specific to his company, he can just add it to the knowledge base, and the AI will learn from it. So if you have your company-specific information, or we noticed that many customers are adding their HR policies or a lot of different information into the knowledge base, and the AI is aware of it.

(Oshri Moyal at 00:07:02) And every time the user is asking some question that the answer is located in the knowledge base, we will gather the information and answer to the users. And this is, actually, if you think about it, I said that it's relevant for CEOs and CIOs. The impact that it has on the entire company is amazing, because we know from our system that, you know, first time to respond to a ticket, the average is three to four hours, which is crazy. If you have a problem, you will wait for IT technicians for three or four hours. And when you have the Autopilot, it's immediate. You get the response immediately, and you can continue. And also, to close a ticket, it usually takes 24 hours on average. And we're seeing a matter of minutes in our system in the Autopilot case. So it has a tremendous impact on the entire company.

(Joel Beasley at 00:08:07) How long ago did you found the company?

(Oshri Moyal at 00:08:10) Wow. It's a good question. I think I had some previous life. We started, I think, 2006 as a security company. We developed the first firewall for the SMB market. It was an amazing and innovative solution, and I think that's what led us to realize and understand the day-to-day struggles that the IT technicians have. And that was the first round of Atera, and it was quite difficult because, you know, creating a firewall is something very difficult, and we have to deal with the hardware. And, you know, when you're dealing with hardware, you need to store it, you need to transfer it from country to country. If you sell in the U.S., you need to take those boxes to the U.S. to ship it and things like that. It's very hard—it was very hard for a small startup as Atera was in 2006.

(Oshri Moyal at 00:09:24) And I think it led us to where we are today because the customers were IT technicians. There wasn't a lot of awareness of security back then. There wasn't security teams in the SMB or CISO or things like that. This happened years after. So the IT technician was our customer, and we realized that once he installed the firewall, he just, you know, threw it into the closet and never saw it again. And we tried to help him much more with his day-to-day, and we saw the struggles, and we saw that he was using five different tools during his day fixing problems. He had the remote systems and monitoring systems and help desk systems, and all of them were legacy solutions, you know, gray menus and very clunky and complicated. And once we saw it, we knew that we needed to shift. And that's what we did. We shifted, and we had this concept of an all-in-one solution.

(Oshri Moyal at 00:10:41) We said we must gather everything together, the ticketing system with the monitoring and remote, and create one streamlined process. You know, they just come to the computer in the morning, seeing the alerts and tickets in one place. The system will prioritize everything, and they will solve a ticket. By the end of the day, it will calculate how much time they worked on this issue. If they are running an MSP business, it will generate an invoice for them. We had to solve it once we saw it.

(Joel Beasley at 00:11:17) That is so cool. So you build this product, this firewall product for SMBs, which gets your foot in the door. So you start learning about that customer, and then they had an expanded problem set and you expanded your business to help them solve that because the hardest thing to do is build those relationships. Right? And have inside information about what the company is actually experiencing. That's a brilliant, brilliant move on your part or your team's part.

(Oshri Moyal at 00:11:45) Yeah. Yep. It was.

(Joel Beasley at 00:11:48) Yeah. And then AI, as we know it today, I'd say maybe two years ago when OpenAI released ChatGPT, that's when it came on the scene for the average person. That's when my grandparents started asking me about AI. Right? So when my grandparents asked me about Bitcoin, I was like, it's time to start selling some Bitcoin. When they asked me about AI, I'm like, alright, it's permeated all the way through society now. And so when did it become clear to you that you could integrate large language models, if you do, into your products?

(Oshri Moyal at 00:12:28) So I mentioned that during 2016, we started hearing about machine learning and big data and everything, and we had this idea that became a patent to solve those problems. I think in 2019, we decided to go and try to implement the solution. So we recruited a bunch of very smart people, you know, data scientists and data analysts and many positions in the company, and we started working on the solution. We knew exactly what we wanted to achieve. And we started working, and it took two years, I think, of very hard work to get to a point. And we felt like it's not exactly like we wanted it to be in terms of technology, in terms of the models that we created. And then one day, we have an amazing relationship with Microsoft, and our system is built on Azure, and we have a very good relationship. They called us and said, we just partnered with OpenAI. We know that you are working on some models. You have to give it a try.

(Oshri Moyal at 00:13:54) And that was way before ChatGPT was released. So we tried that, and we saw immediately that, I think, things that took us one year to develop internally, the OpenAI models did within seconds. So we said, okay. Now we have the infrastructure that we wanted to implement the solution that we have dreamt of since 2016. Let's do it. And we started working on it very intensively, and the results were amazing. Amazing. We were able to do a lot of stuff, a lot of things that we wished the teams here would be able to do. And that made a big difference, and we are now still working with Microsoft very closely on this and continue developing the system. And we have a lot of new innovations and new things that we're continuously adding to the system. So we really believe in large language models, and there is a lot of potential, and we really like it.

(Joel Beasley at 00:15:17) As a CTO, if I'm looking at this and let's say, you know, I got my start in real estate software. So I got to see inside of a bunch of really large real estate companies, and I always interfaced with their help desk to some extent because we'd release a new application into their ecosystem, and then they would get help desk tickets. So I didn't work at the help desk in that sense, but I interfaced with them. And these help desks got pretty big. I mean, there were 20, 50, 100 people just providing IT-type support to, you know, an organization that has tens of thousands or hundreds of thousands of employees. So when I'm a CTO and I'm looking at this technology, there's two things I'm thinking of. The first one is, it's gonna make my relationships better with my executive peers because it's almost always a conversation that's being had that, oh, IT support takes so long, IT support—I didn't visit a company that didn't have that problem. Every company had that tension there. So I see it as a way to relieve the tension and provide a higher quality level of service to our people. That's my first-order effect because that's what I'm thinking about primarily.

(Joel Beasley at 00:16:32) Secondarily, I'm thinking, oh, what am I gonna do when I have a staff of 50 and now 25 of them have nothing to do all day? Do we just reduce and just say that's just what we're doing and, you know, get them help to reallocate them within the org? Or have you seen any of this happen? What type of creative solutions are CTOs doing to handle—it's a rapid-onset efficiency. It's definitely a plus, but it creates these other things you have to think about. How are people handling such a tremendous efficiency when you're implementing the software into their companies?

(Oshri Moyal at 00:17:08) So first, I must say that there is a big shortage in IT people around the U.S., around the globe. They're always needed. And I think what's very nice about it, and we have a lot of conversations with IT technicians and our customers and everyone. They don't like to do those mundane repetitive tasks. They don't like to reset your passwords all day. They don't like to deal with your slow computer, onboard and offboard users, doing the same things every day. They don't like it. They do want to deal with the more complicated, with security, with infrastructure, with tech debt. They want to do much more sophisticated work. They don't want to do the simple one, and they're repetitive.

(Oshri Moyal at 00:18:09) And, of course, for the company, it's, you know, when you're waiting for a technician to answer for a few hours and you can't work and you can't do whatever you need to do, that's a big inefficiency in the entire company, not only in the IT team. So what we see is that the customers that are using our system—it liberates them from those tasks that they don't like to do, and they are using and augmenting the AI to do much more sophisticated work. They are doing a lot of infrastructure and security, and their network is much healthier than others. And I think that's the important thing, and it actually makes them get back to loving to do what they do again. Some of them didn't like it.

(Joel Beasley at 00:19:17) Nobody wants to talk to me about this, by the way. Yeah. Nobody wants to talk to me about the fact that I've talked with so many different people and everybody just doesn't want to have that conversation. And it's tough for me because there are CTOs, and they're like, okay, I'm gonna implement this for Atera or not. They're like, I'm gonna implement this technology, and then now I have—now I'm gonna need a requirement of half the staff or 20% of the staff. And it's like, how do they handle that? Do they just say, we have to press forward? How do you handle that as a technology leader? And yeah, that's the question. How do you handle that?

(Oshri Moyal at 00:20:01) It's a good question. But as I mentioned, I don't see another solution because, you know, progress and AI is there. Companies and competitors will use it. So if you won't do it, somebody else will do it and, you know, you need to progress. That's—I think we've been in—we saw revolutions like that before, and the companies and the managers that didn't embrace the new technology and the new way of doing things got left behind.

(Oshri Moyal at 00:20:34) And some of them I know lost their companies or things like that and became irrelevant.

(Joel Beasley at 00:20:43) Yeah. I think it's an important conversation to have because early in my career, some of the software that I built would displace like half of a department, right? Because it was just so much more efficient. You just needed less people.

(Joel Beasley at 00:20:56) And so every time we went in, it was the same story. Every time we went in, we'd say, okay, we're implementing this new financial—it was a financial software. We're implementing this new financial software. Let's say that there's 10 people on this team.

(Joel Beasley at 00:21:10) Inevitably, eight of them would just complain, be upset about it, not happy. Oh, I gotta learn something. And then two of them would be like, oh, this is so exciting. It's gonna make things so much better. Like, let's go. Like, teach me about this.

(Joel Beasley at 00:21:26) They would sit next to you and learn and help with the whole implementation and ask questions and be engaged. Inevitably, those two people that knew how to operate and run the system the best were the ones who kept their jobs, and the other people got reallocated or left or whatnot. And so I think it's important for people because we're facing this as technologists. You and I aren't okay either. There's gonna be deepfakes for people doing podcasts.

(Joel Beasley at 00:21:53) There's gonna—and so when this technology is seen coming down the pipeline, it's just super important to figure out how to use it and leverage it so that you're not left behind. And that's hard.

(Oshri Moyal at 00:22:07) You know, it is. Absolutely. And I do feel the same thing, you know. I think it was Jensen said that no one needs to learn computer science anymore.

(Oshri Moyal at 00:22:21) So will I lose my job as a developer? I really like to do it, but you know, if I see it today—by the way, I'm using GitHub Copilot, which is amazing. I just write some comment with exactly what I want the code to do, and GitHub Copilot just writes the entire code for me, and it's amazing. You know what? I can sit around and cry and say this thing will replace me.

(Oshri Moyal at 00:22:52) I don't want to use it. But no, I'm using it, and I'm a 10x developer using this technology because, you know, it makes me much more efficient, and I can create a lot of software using it. I don't see any other way. If I won't use it, as you said, someone else will.

(Joel Beasley at 00:23:14) The next generation. The 13-year-old at home with nothing to do after school will just start.

(Oshri Moyal at 00:23:20) Exactly.

(Joel Beasley at 00:23:21) I mean, I was that 13-year-old. I did that. You know?

(Oshri Moyal at 00:23:24) Me too.

(Joel Beasley at 00:23:25) And so now I'm always paying attention. I'm like, alright, what's the person that has unlimited time and no cost? What technologies are they playing with? What ways are they implementing and inventing that's going to put me out of business? Things like that.

(Oshri Moyal at 00:23:45) Absolutely. I'm thinking the same way.

(Joel Beasley at 00:23:49) So what's the thing that you're most excited about that's happening at Atera that you can talk about publicly?

(Oshri Moyal at 00:24:00) Yeah. I think innovation at this point is something that is growing dramatically because, as I mentioned, we had a lot of dreams and things we wanted to fix in this world. We had a meeting, I think it was a year ago. Again, it was a meeting that we are doing from time to time. We call it "Reinventing IT Using the Current Technology."

(Oshri Moyal at 00:24:31) We have this periodic meetings that we are saying, you know what, technology has changed. We have new tools now to work with. How can we make this world better, much more efficient, fun, or things like that? So I think I'm really excited about the innovation that this company is doing and supporting.

(Oshri Moyal at 00:24:56) We're really encouraging people here to innovate and create new things in everything we do. And the fact that it's now possible using the technology, it's amazing. Because, you know, we just started to see new innovations using AI. Companies now just realized that they can use it and leverage it and things like that.

(Oshri Moyal at 00:25:25) We also have this mindset shift, and we're now working on amazing things. I can't reveal everything we do, but we're doing amazing things here.

(Joel Beasley at 00:25:39) In your personal life, you know, not just at Atera, but as you go about in life, what's the coolest thing that you have seen with large language models?

(Oshri Moyal at 00:25:52) Wow. I'm really excited every time I see Midjourney generating images. I don't know why, but art is something that I really love. The level of details, sharpness, you know, capabilities. I'm excited about it.

(Oshri Moyal at 00:26:15) And I know it's one of the first things that happened, but I'm still very excited about it.

(Joel Beasley at 00:26:24) Yeah. It's getting really good. It's getting really good. Okay, I will share this small story with you. Personal story here. So my daughter, she came to me and she said, "Dad, can I start a YouTube channel?" Part of me was like, no. She came to me two years ago and asked this, and I was like, no, you're five, six, or whatever. Like, it's too young, too young.

(Joel Beasley at 00:26:46) And then she came to me again and she's like, "I really wanna do it." And I was like, "Well, what do you wanna do?" And she goes, "Well, I wanna do a channel called 'Sweet Aria,' and I'm gonna go to different shops that sell cookies and stuff, and I'm gonna try them. I'm gonna eat sweets and try them." And I was like, what a what a seven-year-old show idea, right? They want more treats all the time. It's like, I'm gonna make a show that does this. And so I went to ChatGPT, and I said, "Hey, ChatGPT, my daughter, she wants this channel." I described it. I said, "Can you please generate a logo and a YouTube image for her?" Let me show you.

(Oshri Moyal at 00:27:28) Amazing. Amazing.

(Joel Beasley at 00:27:30) Wow. It was amazing.

(Oshri Moyal at 00:27:32) I can generate images all day long. That's crazy.

(Joel Beasley at 00:27:37) Yeah. And it had the—it's figured out the text now, because two years ago, it was like just putting random characters. But it's figured out the text. I was just like, this is unbelievable. This is crazy.

(Oshri Moyal at 00:27:49) How much time you had to spend if you do it five years ago or three years ago?

(Joel Beasley at 00:27:55) Three years ago, I would have written up a design brief that was more in-depth than the two or three sentences I used. I would have to go collect examples of what I wanted it to look like because designers need examples of your style that you like. It was all of this work, conversation back and forth. If I like this, I don't like—but this was not that. And it was amazing.

(Oshri Moyal at 00:28:19) It's crazy. That's exactly the process. So we had, I think three years ago, we had a hackathon in a hotel, and I created very fast the system that does phishing attacks for learning purposes. And we wanted to create a logo and some banners and things like that.

(Oshri Moyal at 00:28:43) So I did exactly what you said: brief and, you know, contacted some designers in Upwork, I think it was. And think about it. If I did it today, I was doing it today, it would take me seconds. And I spent, I think, hours on this.

(Joel Beasley at 00:29:04) Amazing. So let's wrap up with the calls to action for Atera. For the people that are experiencing—either they can't hire enough technicians, help desk technicians, right? Because you said they're limited. Or there are people who are overwhelmed and maybe disillusioned with just doing these repetitive tasks, and you might be able to implement the solution and help them. What do they do? Do they just go to atera.com? Like, how do they get more information?

(Oshri Moyal at 00:29:31) Yeah. They will go to our website, atera.com. They have a 30-day trial, so they can test everything that I mentioned by themselves, experience the AI, and I'll let it do the magic. And by the end of the day, if they are interested, just add their credit card, and they are good to go.

(Joel Beasley at 00:29:55) Nice. That's easy.

(Oshri Moyal at 00:29:57) Yeah.

(Joel Beasley at 00:29:57) Yeah. I like the way you run your business. Make it really easy. Okay. Alright. I've got one last question that I would like you to help me fill a knowledge gap that I have. I saw you wrote some content on AI hallucinations.

(Oshri Moyal at 00:30:13) Mhmm.

(Joel Beasley at 00:30:14) I am not an AI expert, so I don't have a deep understanding of the programmatic inner workings of the AI. I just have some basic understandings. But my question is, what's the difference between a hallucination and just a standard response?

(Oshri Moyal at 00:30:33) As we see it—and we created amazing systems that, you know, run IT problems over and over again, changing the way the user writes the text. You know, we're using also AI to create those kind of tests. And we're seeing that in some cases where the AI doesn't have the answer, it's just inventing something which does not exist. And if you think about it, large language models are just trying to predict the next word after a few words that you enter. It's called completion.

(Oshri Moyal at 00:31:22) So if they search, you know, the entire data that they are filled with, they will find something to say. It will have lower probability, but they will find. And this is why the hallucinations happen. We dealt with this problem in the past two years with a lot of consultants and Microsoft teams, and we found a way that after a while, we realized that we are doing something that we did many years before, you know, trying to teach the AI to say, "I don't know." I'm trying to teach my engineers to say, "I don't know" for 10 years now.

(Joel Beasley at 00:32:23) So we—

(Oshri Moyal at 00:32:24) Good luck. We found a solution also for that. And yeah, it's a continuous work. We are dealing with that. It's an issue. We know it. But there are some techniques, and Microsoft really helped us a lot with this issue.

(Joel Beasley at 00:32:43) Oh, okay. So programmatically, you can tell when it doesn't know, when it's going to make something up. And that's a different signal than when it does know and it's just telling you based on the data.

(Oshri Moyal at 00:32:54) We find a way to know when it's making up and, you know, double-check it and things like that.

(Joel Beasley at 00:33:03) And then, because the audience is mostly tech leaders, can you just share one story of a hard lesson that you've learned in leadership while building this company?

(Oshri Moyal at 00:33:13) Wow. I think there are a lot of stories. I think maybe it's obvious, but the company has grown, you know, from a startup in the garage with six people to almost 500 employees. And there is, I think, one point in time that you realize that the company is growing, and you need to start working as much as you work hard on, you know, developing something or sales or marketing. You need to start working very hard to keep your culture and how to work on making sure that the people around you will be the people you want to be with during the day in the same room, and the atmosphere will be amazing and good and fun.

(Oshri Moyal at 00:34:14) And I think one of the core values of Atera and something that if you ask anyone in the hall is enjoying the ride and work as a team, and these values are great. And I'm quite sure that it's also part of the reasons why the company is growing and doing amazing work and succeeding.

(Joel Beasley at 00:34:42) Well, we did it. We made a podcast. Go to atera.com and spend your money with them. Hand them your money. I love what you're doing. It's one of the easiest business sells of all time because you're helping people that have a shortage being hired and helping people that are talented doing repetitive things that they don't enjoy doing in a department of the company that is critical. And so it's an easy, easy sell. It makes sense.

(Oshri Moyal at 00:35:12) Absolutely. Well, thank you so much. It was amazing. Thank you.

(Joel Beasley at 00:35:17) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email: [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.