Beyond AI Risk: How Bonfy.AI Is Securing Data for the Future

Hey, everyone. Welcome back here to Techstrong TV for another interview. I am really happy to have this gentleman on.

You know, I, I first met Gidi Cohen. I'm going to guess it was 15 years ago. Giddy.

Yeah. Maybe More, at least. Yeah, at least.

Uh, Giddy had I, I think it time Giddy had just recently come out of IDF or had, he was starting a company, right? Called Skybox Security. And he showed me something that at the time rocked my world, right?

It was, it was called an, like an attack map, a 3D attack map, where, where it actually, you could, it looked at your network infrastructure, your vulnerability profile, and could draw a map of how a hacker would attack you and what you can do to prevent it at various places throughout your network. Sounds matter of fact, now I still see Giddy, I still see companies showing me that today, Right? Yeah.

Like came, like, came just out Ofir. We, and I asked them, I, we Read that 20 years ago. Yes, I seen about a guy named Giddy.

They're like, no, no. I said, I saw this 20 years ago. Finally, it's funny, just a, a couple days ago, I, I was talking to a founder of a company, an Israeli company, and, and he was talking, I said, you know, this sounds a lot like a company I knew called Sky.

Sky. He said, Skybox. I said, yes.

He said, yeah, I know. We, we, I, I remember much younger I was using that. And anyway, it was good stuff.

Giddy has been a, a pioneer in, in what today we call Cyber for a long time. He has a new company. He's, he's co-founded called Bonfire.

We're gonna get into, but let's welcome Giddy, welcome to Techstrong tv. It's a pleasure to have you on. Yeah, thanks.

So I didn't mean to my pleasure you, and give you the whole like, background, but I, I didn't give you the whole story. Why don't you share with the people a little bit about your story? Sure.

So let's start with actually now, and let's go back a bit. So, I live in the Silicon Valley in Palo Alto for quite many years. That's where we founded Phi ai.

We're going to talk about, uh, soon as well. But background actually started, uh, with IDF, as you mentioned, uh, back in Israel in 8,200, right? The, like the Israeli NSA, but I mean, kind of thing.

The, all of my life was in the cyber data analytics technology since then. And they seen and um, and basically now bonfire, right? My third company, my third startup starting, all of them are dealing with different facets.

Let's call that way of cyber data analytics. And now all of that they place together. Uh, so that's very absolutely, that's my very, very short deck on This gi I I've started a few companies.

You have, it's not something you do, you know, lightly. It's something you make a commitment to. You have to be passionate.

Yeah. You gotta believe in what you're doing. You gotta think that in some, at least even a small way, what you're doing makes the world better.

It makes, there's a reason people want, need to have what you're, you're developing. Talk to me about your passion for bonfire. Why, why bonfire?

What was the passion that drove you for this? Yeah, so, uh, Danny Bel, my co-founder and CTO and myself, uh, got together almost two years ago. And, uh, I, it was just after I left Skybox, uh, about six months before, nine months before Danny left cyber rock, he managed the entire RD of Cyber rock.

And he said, okay, we, we see what's going on with ai. That's just the beginning, not of ai. Of course that started, yes, between us 30 years ago, but the massive adoption with, of gen AI started the end of 2022, beginning of 2023.

And we said, that's going to be great opportunity for innovation on one end, right? Adopting smart beast, uh, called land for a lot of different uses, but we believe that can create a huge amount of risk for people, society, organizations. So both then is a, a nice a mining background, right?

Are in enterprise cyber security. So we said, okay, we want to tackle, uh, all of the challenges and help organizations adopt AI to do it much more safely, trustworthy, without, uh, all of the exposures to everyone could just imagine. Then I think there are a lot of proof proofs since then the, those exposure really happened.

So that's, that's, we said, okay, we must be part of that, right? We are technologies, we love cyber and AI and data now, machine learning and the, all of those new technologies with ai and we want to put it together to work for the benefit of, uh, enterprise around the globe. And that's how we started when we dug in More than that, it that there are probably two, and maybe there would be more along the way, but two main areas from cybersecurity.

One, you can either think about, let's deal with the model, let's protect the model, prompt response, supply chain vulnerabilities, a lot of other stuff. We've seen a lot of startups do that. We decide not to go on this route, not because it's not viable, but because it's, we thought one, it's crowded and I think it was proven correct, and second, the we're not sure.

That's where the majority of the focus of enterprises as they adopt gen AI will happen. Therefore, we, we decide to go different route, which focusing on the data side of that, right? At the end of the day, the entire reason cybersecurity exists, the, the things you feed models with is information or data.

They feed with data. You get data out of that. It's true regardless of the LLM is, uh, used by individual or embedded in the system like Microsoft 365 copilot and for the benefit for productivity of users, whatever form factor, it's all about the data, data that gets into AI or AI enabled machines, data that comes out of that with humans by agents or whatever it might be.

So we said to ourself, that's a huge problem to solve and that's what we want to tackle. Maybe last piece of the, uh, origination of, uh, bonafide is that, um, which by the way, bonfire stands for IDE ai, if I didn't say that before. So AI in good faith.

So the other fest of that, it was clear to us that why AI creates a lot of new challenges, new exposures to organizations, which is, which are dangerous, risky, and is a vendor exciting to solve and help organizations deal with it. Actually, there's also a huge amount of problems that print data, gene ai, data security, and for unstructured data, uh, one of the spaces that I would say was the least serve with quality technologies over the last 15, 20 years. Everyone knows about DLPs and false positives and false negatives and total cost of ownership, which is all true, and all of that is now getting much tougher when you are putting this, uh, ai, right, ML gen, ai, AI agent on top of that, that we say, okay, that's exciting problem as a startup, right?

To address both addressing the, the rising issues, uh, and the expansion of let's go the data surface, but also with the same soup, basically handle a lot of those historical challenges that were never sold properly. And we thought that's a great opportunity to open the prices. That's, and that's why we started the company.

Absolutely. You know, it, it's a, it's a common thing in security. Giddy.

If we could just stop the world for a day or two and let us catch up, then we'd be okay. But unfortunately in security, it seems like we're always, we're always trying to catch up. We're always trying to catch up and, and new stuff keeps piling in.

Yes. So you, you look at something like legacy DLP, right? You and I have seen DLP solutions come and go and yes, we never quite got it right.

And now, you know, you start adding AI and, and the, the, and the, the, the sheer volume of data, let alone the velocity and the automation. There's no human in the loop sometimes with this stuff. Exactly.

If we didn't get it right before, how, you know, how are you going to get it right now? But every once in a while technology affords you a a little bit of a miracle, right? The technology allows you to almost make up for past mistakes because with this newer technology, not only am I able to work with things like ai, but it allows me to maybe hit the legacy DLP issue square in the head too.

And I I, you know, I think exactly, exactly. That's the opportunity here for bonfire. Yeah.

Uh, absolutely. Right. So I would say that from innovation perspective, which is that's what get me excited, uh, excited, right, as an entrepreneur, right, is that, uh, we, we actually take advantage of two things and bring it together in technology we developed, okay, which is now, by the way, available in production and, uh, anyone wants to see it, we'd love to people to get, get to us, send me an email, get to our website, et cetera.

But there are two concepts we actually brought together, or two abilities. One, obviously, right? With all of the innovation of Gene ai, not only create risk, but also create, create set of tools and technology we could use.

And of course, we're using, you know, to understand content in much better ways for the benefit of classification, content analysis, detection, prevention, and lot of other use cases that are related to data security. So that's one, but that's not enough. And you mentioned Skybox Security, right?

By previous company. They ran for a, a was one of the founders ran for many years. One of the innovations we had there in day one, as you mentioned, is how to utilize Netto context on the infrastructure in order to manage vulnerabilities better and manage compliance better, et cetera.

So that's, that's how we actually start the company use, using context for the benefit of, uh, various use cases and provide a lot more intelligence, uh, and accuracy, therefore, uh, for, uh, for those type use cases. So think about what we did in what we're doing. Bon is actually the analogy, nothing to do with the technology of skywalk, right?

We're not dealing with infrastructure there, we're dealing with data and content information, but actually the innovation we have here is actually how to utilize business context that we serve, learn from the organization perspective, their own data for their benefits, solely. Of course, we never take the data. Other customers never see that.

So it's all securely done for them and how to utilize this business context for the benefit of analyzing their data in motion, addressed, generated by machines, generated by humans, all of those type of flows. And that's basically what bring together the kind of modern AI technologies we are developing and using, plus the innovation of how to utilize smartly business context, put them together to significantly smarter, uh, datasecurity solution, which is significantly more accurate. Covers, uh, probably five times more scenarios in terms of, uh, uh, the things you want to detect, which are critical for this modern world of, uh, of, uh, data security or data compliance type of risks.

And to do it in a much more cost effective way in terms of the amount of human load on the security team set up dealing with the operation matters significantly more streamlined than ever Before. Excellent. I love it.

Um, you know what, let me do some housekeeping. Giddy Bon fee. Bon bonfire is B.

Yes. BON. Yes.

ai is the website Do ai And then the, the, the product service, if you will, is bonfire's Adaptive content security. Yeah. Or bonfire a CS.

Exactly. This our platform. Exactly.

How could people check it out, test it out? Is there a free version of trial version, something they could get their hands on? Yeah.

Then no, that's, that's great. So first of all, the best and the easiest way is contact us via the website. There are multiple forms for different type of interest, uh, the, that the customers prospects, uh, might want to have.

We'll be happy to provide demos, uh, schedule some qvs right value so we can actually demonstrate for organizations that can walk in their environment. It's super simple. We design part of our lessons, right?

All of us, right? Lesson serving and enterprise cybersecurity users and buyers along the years that, uh, on one end, right, the industry grew bigger and bigger, but there are more solutions and the load and securities is actually, some of it is un verbal, I would say, in terms of complexity of solutions and the amount of knowledge they have. So bonfire would design on one end, super sophisticated backend business context, learning other stuff for the benefit of the use case, but very easy to deploy, very to, to use.

So that's true for production, but also very easy to perform a put value with our solution. So just contact us and we, we'll be happy to show it and, and show ly our talks in your environment. Yeah.

Giddy, I, I, I speak to a lot of security folks, a lot of entrepreneurs, ai, DevOps, platform engineers, and, and I think the consensus is, look, you are going to need AI to secure ai. You're going to need AI security solutions to defend Yeah, for sure. AI malware or AI assisted malware and and so forth.

If you wouldn't mind, talk a little bit about how Bonfire is using AI to make this solution better. Yeah, so we are, we are using, uh, multiple, actually both models and AI technologies, as you can imagine, uh, at bonfire, right? To provide the answer at scale.

Inaccurately requires a lot of technology. It's just, it's not one thing, right? And if someone think that, yeah, I'm just going to get, let's say the email, I want to analyze and send to the DLP, the, to the LLM as a DLP function to tell me if it's good or not, that wouldn't work.

So we we're mixing together multiple technologies of that. We develop something that we are taking off the shelf and tuning for our needs. So for example, knowledge graph creation, the business context entity linking technologies, so can identify what the entities in the content.

Uh, we have different AI engine engines that are looking for features in the content that allows us to assemble together to understand what's there. So we can, in a very abstract way, understand the, you know, we would've discounted this, uh, good or not, right to pass through an email or file sharing or web traffic or whatever it might be. So we, we have that capability.

We have a deep explainability capability. So someone gets an alert. Part of the issue once they get alert, of course, you want it to be accurate, right?

But you want it to be understood, right? Think about the, a security analyst, get an alert from Bon ffi or any other system, something might have happened. They have no context what it is, right?

How do you know that? How do you help that? So we have AI used for explainability of a, of a, let's say, complex alerts.

So we are using it in, in, in quite a few different ways because it's necessary, right? And that's part of our technology, software developed by ourselves, as I said, some that we're adapting existing models for, for the need. Small models, mid-size models, large models, and plus our knowledge graph and learning technology and all this is put together in a solution.

Jenny, We're about out time. I just wanna emphasize this is available now. People go to the website right now and go check it out.

Yeah. We're in production already serving customers, looking for more customers as we build up and penetrate the market. Uh, yes, gen, please call it US Gen here.

Giddy, you're out in front. Again, mazel tough as they say, right? Good for you.

Keep us posted, come back. Yes. Keep us posted on progress here and what you're seeing.

Absolutely. I'm happy to be there anytime you Invite me. Gih, uh, co-founder, CEO of bony, that's B-O-N-F-Y, ai bony, excuse me, me Bony.

You know, I got this funny French accent bonfire. Yeah. All of us on accents.

Yes. Go. Yeah.

For, for letting speakers, letting speakers can say b****s. It's Fine. Come from the Website for letting speakers or Latin languages Bonfire ai.

Exactly. It, it's something that's needed in today's ai. We'll check it out.

We're gonna take a break here on Text Drunk tv. We'll be back in a minute.