The Future of AI in Data Protection: A Conversation with Alcion CEO Niraj Tolia

[00:00:02] So how are new companies thinking about the way they're going to infuse AI into

[00:00:07] products? And what's the tension between older legacy companies and the new

[00:00:13] startups nipping at their heels? Well, Nourish Tala joins me.

[00:00:16] He's the CEO of Alcyon who focuses on data protection from Microsoft 365 on this

[00:00:21] bonus episode of the Business of Tech.

[00:00:25] Are you ready to spot opportunities by aligning IT with your clients' business

[00:00:29] strategies? Get in Sync equips MSPs and IT professionals with the tools, methods,

[00:00:35] and training to deeply understand client strategies, ensuring IT investments

[00:00:40] directly support key business objectives for tangible outcomes. With Get in Sync,

[00:00:45] you gain critical insights that empower decisive actions, enhancing your

[00:00:50] competitive offering. This solidifies your role as a trusted advisor and

[00:00:54] supports your clients' strategic needs, bringing greater success. Test your

[00:00:59] readiness to become a certified Get in Sync trusted business advisor with our

[00:01:04] free online assessment. Accept the challenge and discover if you have what

[00:01:08] it takes to become an indispensable strategic partner for your clients. Begin

[00:01:13] your journey with Get in Sync. Visit getinsync.ca slash MSP Radio to learn

[00:01:20] more.

[00:01:23] Nourish, thanks for joining me today.

[00:01:25] I'm very excited to be talking to you today. Thank you for having me.

[00:01:29] Me too, because you know, one of the things I keep talking about on the show

[00:01:32] all the time is all of the various approaches that around AI and what we're

[00:01:36] seeing. And obviously, that's everyone's buzz. But what I think is interesting

[00:01:40] is, is that there's lots of different approaches about making AI specific,

[00:01:46] similar, you know, our audience of IT services companies, MSPs, and how we

[00:01:51] translate that into what users do. Tell me how you specifically think AI is

[00:01:57] important, is going to impact this space?

[00:02:00] Sure. So, you know, my vision about AI isn't in particular for the channel for

[00:02:07] MSPs has always been the same over the last few years that we've been doing

[00:02:11] this. But you know, when I rewind the clock a year ago, I think there was some

[00:02:14] concern on the channel, is AI going to replace us? Is AI going to impact our

[00:02:19] business, etc? But that's on the way. To answer your question, I think AI is an

[00:02:24] amazing opportunity for the channel, the MSP, for the service provider out there.

[00:02:29] Right? Because the way we look at it, the modern MSPs, what they're trying to say

[00:02:35] is, how do I replace my legacy tools with AI enabled tools so that I can do a

[00:02:43] better job and focus on the things that I really care about, which is my end

[00:02:46] customers, right? My professional services, their digital transformation.

[00:02:51] Because AI, it is giving me new features, but it's taking away operational

[00:02:56] quality from me. And so the way I look at our channel partners today, we're at

[00:03:01] the forefront of implementing AI. It is about how do I both deliver a better

[00:03:08] experience for the end user, elevate my position and reduce all my operational

[00:03:12] toil, but to focus on those value added things that I, right? And that is the

[00:03:17] consistent message we've been hearing over and over again.

[00:03:22] Now you said something there I'm really curious about. So you talked about

[00:03:25] replacing legacy tools with new tools that are AI enabled. Yes. If I were to

[00:03:30] take a parallel to the larger tech companies, right, those legacy players

[00:03:36] are the ones that are generally bringing AI to market. So for example, we think

[00:03:40] AI, we're thinking for soft and Google and Amazon, who I would not say are old

[00:03:46] players. So there would be a solid argument for most providers to say,

[00:03:50] instead, my existing toolkit is probably going to get an upgrade. And it sounds

[00:03:55] like you've got a different opinion. Tell me what your thinking is of that

[00:03:59] legacy versus new and the challenges.

[00:04:03] I think, drawing another analogy, AI in some sense today is like the gold rush,

[00:04:07] right? When I go look at Google, Microsoft, what they're doing with open

[00:04:11] AI as an example, the creator of chat GPT. Those are the people that are

[00:04:16] providing the shots, right? And the pick access to say, look, you're the tools,

[00:04:22] but are they readily, are these the things that would be directly consumed

[00:04:27] by the IT practitioner? No. Right. Sometimes the answer is, you know, maybe

[00:04:33] a little bit, right. I take care of a translation or some support requires. So

[00:04:37] pull these out of my knowledge base, but usually they're not directly

[00:04:41] implemented. Right. When we go look at any of our partners today, all of them

[00:04:46] have a stack of tools, right? For backup or memory for security as an example.

[00:04:53] And there's a lot of quote unquote AI washing going on where people will say,

[00:04:56] look, I do some statistics. It's AI. I do AI, but it's really, it's a demo

[00:05:01] quality version of that. I think to rate for what we've done at Alicent and

[00:05:06] we'll talk a little bit more about that later. We said, starting from a blank

[00:05:09] slate, if we were to build this on the ground up, or there's a reason that's

[00:05:14] AI in our domain name and our company name, we did this on the ground up.

[00:05:17] What would the experience, what are the features that we can provide? And some

[00:05:22] of this stuff then becomes a cost issue, right? How can we provide cost effectively

[00:05:26] people? And you can't do it by slap bolting AI on, right? Do a legacy product.

[00:05:33] I think the bigger companies are providing the basic tooling, the

[00:05:38] foundations, I would say. The house we construct to solve our channel partner

[00:05:43] problems is actually going to be very different. Right? So it's like building

[00:05:48] a house now versus building a house or ranch, you know, a hundred years from

[00:05:51] now, construction techniques have changed a lot. And that's the analogy I would

[00:05:57] draw here. That is, I would answer yes, but doing it, applying AI in a domain

[00:06:03] specific manner is a lot more involved than saying, Hey, you know, let's add an

[00:06:10] LLM to my product and be done.

[00:06:13] So make that a little specific for me in the world of backup, because, you know,

[00:06:17] this is a certainly an arguable, very mature space, right? We know how to

[00:06:23] generally backup data. What does AI bring to the table when you infuse it into

[00:06:29] backup technology?

[00:06:30] So right for your listeners, I do want to give some context, right? In terms of a

[00:06:34] little bit of my background history, I am very passionate about data protection,

[00:06:38] right? I did my PhD on the subject and I've been through many companies, right?

[00:06:41] And many startups. The first startup where I led engineering got acquired by

[00:06:45] Dell EMC, right? Company Advice got acquired by Palmwall. And then company I

[00:06:51] founded next got acquired by Wien and on to my next company now at Alstom.

[00:06:56] BearBeam is actually one for larger investors. All right. They let us do

[00:07:00] resale. And so I've been in data protection for a while. And, you know, I

[00:07:04] thought I was done with founding companies, right? To be very blunt. And

[00:07:09] that's when we started running into what AI could do, how the climate has

[00:07:14] changed. And then particularly thing, right? We said, why is data loss going

[00:07:19] to come, right? It's no longer my district failing, right? Yes, they fail.

[00:07:23] And we need against that. But it's really this whole cyber threat climate

[00:07:27] we live in. Malware, ransomware, right? And we see every business doesn't matter

[00:07:33] whether it's small or large getting impacted. Literally see that small 5

[00:07:37] person companies, large 5000 person companies all impact that. And the only

[00:07:42] real way to tackle those issues in data protection is using AI. And that's why

[00:07:47] we founded Alstom to do it. Right? So hopefully that gives you a little bit

[00:07:51] more, I can go into a lot more depth. But when we look at just how the

[00:07:55] landscape has changed in terms of tag vectors, in terms of what people are

[00:08:00] out there doing in terms of movement to the cloud and a more collaborative way

[00:08:04] that companies are working, that's been somewhat of a very concerning

[00:08:10] confluence of events that led us to say, look, we need to go and resolve this

[00:08:16] data protection problem, because the legacy we have doing it is not going to

[00:08:19] work anymore.

[00:08:21] So what does AI bring, though? Because I could if I remove AI from everything

[00:08:25] you just said, it's still all internally coherent, right? Like, so, so what does

[00:08:29] AI bring that makes it different in this?

[00:08:32] So let's see, there are, it brings two or three things, right? So it brings

[00:08:37] from the let's talk about it first from the end customer perspective, right?

[00:08:41] What the end customers get a number of great features, including improved

[00:08:45] security, you can't do ransomware manually, right? You get improved.

[00:08:51] It's within that your ransomware detection, malware detection, anomaly

[00:08:55] detection, right? Someone suddenly doing something strange, deleting files

[00:08:58] with it in, moving files out, right? There's a lot of work that goes in there,

[00:09:02] right? And for today, for example, we have multiple AI models that are

[00:09:06] continuously learning on every backup what novel behavior is, right? And none

[00:09:11] of that would be possible in a legacy world with heuristical rules.

[00:09:15] But I think it is also more important for our partners, what AI brings to

[00:09:19] for them to the table. They are getting a material reduction in operational

[00:09:25] time, right? We've had so many, and this is heartwarming to me, we've had so

[00:09:30] many partners come up and thank us and say, look, you are literally saving

[00:09:34] hours of my week that I used to be doing in total great fix on my backup

[00:09:41] system. Why? Because on our side, what shows up as ease of use to them is

[00:09:45] AI, we do better scheduling. We wait in a thousand percent coffee. You can

[00:09:50] never set up a manual schedule for person, right? To back up when they're

[00:09:54] the most active and when they modify the most data. Our system can do it for

[00:09:58] our customers, right? You don't need to worry about scheduling anymore. You

[00:10:01] don't need to worry about reliability. We figure out when the best time to do

[00:10:04] it is when to retry something out of auto-remediate some issues, right? So

[00:10:09] from an ease of use and operational perspective, so what suddenly goes away

[00:10:16] that again, you focus on the thing that really bring value to the business and

[00:10:19] not the things you need to break fix internally. And I think those are the two

[00:10:23] big things. When we talk about ease of use, when we talk about reduction in

[00:10:27] pain, when we talk about customer benefit, that is the consistent message

[00:10:31] we have received from all our partners.

[00:10:34] How do you think about the balance of ease of use versus the ability to

[00:10:40] provide insight into what's happening? Because oftentimes when you're doing,

[00:10:44] you're making changes to make things easier to do. You're removing some of

[00:10:47] the controls or some of the exposure to that can be a good thing, but there's a

[00:10:52] balance to be struck there. How do you think about that balance?

[00:10:55] So I think it is not an either or is what we would like, right? So I think

[00:11:02] it is what you present at the top level, how do you go down and how do you not

[00:11:07] overburden someone with complexity to begin with for day-to-day jobs? And so I

[00:11:12] think there's different degrees of complexity, but then now this is going to

[00:11:17] start like I have a hammer and I see a nail and I'm going to use the AI hammer

[00:11:21] and every nail, but this is actually true, right? Some of the things that we

[00:11:24] are internally be looking at is how do you give that control back when you need

[00:11:29] it? Using tools because sometimes I think complexity in legacy products is

[00:11:34] the right because there was use case X, there was use case Y, right? Use case X

[00:11:39] only applied to 5% of users, but they really needed it. And people threw

[00:11:43] everything in, including the kitchen sink. What we have figured out is some of

[00:11:48] these things when we go look at how we use chat GPT, which is talk to mine data

[00:11:52] or talk to scrape data as an example of oversimplifying. We're seeing more newer

[00:11:57] use cases that we can apply even for API, which is chat with my API or chat

[00:12:02] with my data. So the type of use case that you talked about can be readily

[00:12:06] serviced by you can just say, Hey, right? What is happening here? Can you give me

[00:12:10] more insights there? And we have seen how systems like chat GPT with a few minor

[00:12:14] additions can go talk to our APIs and over time go talk to our data, all these

[00:12:19] insights up. So you don't have to build for the entire universe and burden of

[00:12:26] legacy system, burden a system with complexity. You can have it be a little

[00:12:31] bit more freewheeling, which is usually what customers want because most of

[00:12:36] these use cases are very one-off. A customer request comes into a part and

[00:12:39] they say, How do I figure this out?

[00:12:41] Your statements there implied some of where my next question's going. And I

[00:12:45] want to get a little bit more of your thinking of it because when you talk

[00:12:48] about allowing chat GPT to call into APIs and such, it means you've given some

[00:12:52] thought to the way leaders can optimize their relationships, add value to data.

[00:12:57] Give me a sense of your philosophy around it.

[00:13:01] Sure. Right. So first of all, I think this is what you were thinking about. The

[00:13:06] first thing that comes up to us is privacy and security, actually. Right.

[00:13:11] All of these features are great, but if I am leaking data accidentally across

[00:13:15] customers or I have an LLM that exposes data, right? None of those things are

[00:13:21] great for anyone. And us, the partner, the customer, right? So we actually,

[00:13:25] our vision is we started with the foundation of security. We started with

[00:13:29] foundational privacy. So when we train our model, all the models are customer

[00:13:34] specific and nothing's ever shared as an example. Right. Then what is the next

[00:13:40] step up? Then there is what do we do? Use those models only to provide

[00:13:44] features that just come out of the box. They don't get it. The next level up is

[00:13:48] when you think about, what do you do from an API programmatic lens? So for

[00:13:53] example, today we have a customer dashboard in the system. We have a partner

[00:13:56] dashboard in the system. But what if there's a new use case? Right. It's all

[00:14:00] under the customer and partner dashboard. It's the same APIs. So then how do you

[00:14:04] expose a programmatic interface that allows AI related tools to interact with

[00:14:09] APIs? And then finally, and we don't have this yet, is looking at just how do you

[00:14:15] talk to your data? How can you pull up more insight of that? We have businesses

[00:14:19] that said, Neeraj, back up all my critical data. And those are the things you can

[00:14:24] get insights. And how do you then go build that out is what we're looking at,

[00:14:28] is one of the things we're looking to do next.

[00:14:31] Yeah. Isn't there a lot of some work required by the provider and the MSP to

[00:14:35] get data in some kind of readier state? Most organizations are pretty sloppy when

[00:14:42] it comes to data. Isn't there a process required? Tell me about what's required

[00:14:46] to make that use. Sure. So I think there are two ways of looking at it. There is

[00:14:51] a dedicated business case that one might have to say, look, this is critical to my

[00:14:57] business, right? And I'm going to build out some custom tools. I have a full

[00:15:01] pipeline. I'm going to clean the data, ingest it, right. And feed it into whatever

[00:15:06] system I'm building using the tools provided by some of the big cloud providers

[00:15:09] you mentioned earlier. Right. What we're saying is, look, even before business

[00:15:14] case access, what can one do here? Right. Because there is still in some cases,

[00:15:22] the 80-20 kind of rule where you'll get 80% of the benefits with maybe 20% of the

[00:15:29] work done here. And we've already done the heavy lifting of pulling data into our

[00:15:33] system. So how are we then going to expose it when it comes to the other thing

[00:15:37] what can do there? And then once you at least get some insights, you can then

[00:15:41] figure out is it worth the investment for me to do something more fuller below.

[00:15:44] Right. So we're not saying outside is the end all be all of these. All right.

[00:15:49] But we're saying it is that one critical building block that even trusted already

[00:15:52] a critical data to and we can already do intelligent things with that. Right.

[00:15:56] Let's focus on the data protection side. We'll be doing more. And so this becomes

[00:16:00] another again, value add feature for our partners, for our end customers,

[00:16:05] for their end customers. Sorry. Well, I think that's the area where there's value for providers.

[00:16:10] So yeah, Ross Tolia is the CEO of Alcyon focused on the security challenge of data protection

[00:16:15] for Microsoft 365. He's a former Veeam and Dell EMC exec. He's grown the organization

[00:16:20] of customers in 15 countries over six continents and just launched their MSP program. Ross,

[00:16:26] thanks for joining me today. Dave, thank you so much for having me.

[00:16:29] Are you and your clients tired of the time consuming ticket tennis of coordinating meetings

[00:16:36] and help desk calls? Wouldn't it be better to automate this process with a tool that connects

[00:16:42] directly to connect wise manage or auto task. Time zest offers scheduling automation that gives you

[00:16:50] complete control of your schedule and eliminates the hassle of calendar ping pong as the only

[00:16:56] service designed specifically for MSPs. It integrates into your workflow and makes

[00:17:02] scheduling appointments easy on you and your clients. Plus you can try time zest for free.

[00:17:10] Visit times s.com slash MSB radio and use the code MSB radio to get 10% off your first year

[00:17:19] of times. The Business of Tech is written and produced by me, Dave Sobel, under ethics guidelines

[00:17:26] posted at business of.tech. If you like the content, please make sure to hit that like button

[00:17:33] and follow or subscribe. It's free and easy and the best way to support the show and help us grow.

[00:17:40] You can also check out our Patreon where you can join the Business of Tech community

[00:17:45] at patreon.com slash MSB radio or buy our Why do we care merch at business of.tech.

[00:17:54] Finally, if you're interested in advertising on the show, visit MSB radio.com slash engage.

[00:18:01] Once again, thanks for listening to me. I will talk to you again

[00:18:04] on our next episode of the Business of Tech. Part of the MSP radio network.