Kristin Myers: Now companies are putting AI agents to work, software that doesn’t just answer questions but takes action on its own. Now, those agents can open files, change settings, and move data, often for hours with no one watching. Now, my guest spent years on offense in Israel’s elite Unit 8200, learning exactly how systems get broken. And now she’s building the guardrails that keep AI agents in bounds. And today her company is coming out of stealth with $50 million behind it. But first, let’s take a look at the company. The road from an idea to the opening bell starts right now.
Companies are racing to put AI to work. But what happens when an AI agent gets access to sensitive data and then starts making decisions on its own? That’s the risk Yuval Zacharia is going after, a former elite Israeli intelligence officer who later led security research in AI at Hunters. She’s now the co-founder and CEO of Bay. The company is building a security platform that watches what AI agents do and blocks dangerous actions before they happen. The idea: let companies move faster with AI without giving it unlimited control. And as businesses deploy more autonomous AI, that security challenge is only getting bigger. But here’s the real test. Can they become the security layer every company needs before trusting AI to act on its own? From Startup to Stock Exchange, Bay co-founder and CEO Yuval Zacharia is next.
And joining us now is Yuval Zacharia, co-founder and CEO of Bay. Yuval, thank you so much for joining us today.
Yuval Zacharia: Thank you for having me. I’m so excited to be here.
Kristin Myers: Yes, welcome to the floor of the New York Stock Exchange. So let’s talk a little bit about where Bay is now. It’s coming out of stealth today, $50 million in new funding. Congratulations. So tell us a little bit about Bay and what it is you would like to announce today.
Yuval Zacharia: Absolutely. So yes, today Bay’s coming out of stealth officially with $50 million in funding. And we’re here to enable the future of work, right? Organizations, enterprises are working completely differently than what they used to, not so long ago. Right? So we used to operate our own computers, our own systems, but now agents are doing it on our behalf, and that’s exciting, but also in some ways a little bit scary.
Kristin Myers: And that’s actually, you know, a little bit behind the reason your company was founded. So when was it, or what was the moment that you realized that Bay needed to exist?
Yuval Zacharia: So we founded Bay in January. The truth is that we were working on a completely different idea.
Kristin Myers: I’m so sorry. Just to be clear on when?
Yuval Zacharia: January 2026.
Kristin Myers: That’s right. This year, 11 months later, already $50 million in funding. Okay, continue. That’s an incredible trajectory.
Yuval Zacharia: Thank you so much. And I must admit that it’s all thanks to our amazing customers and the amazing traction that we’re getting from the market. We have customers from the Fortune 100 list. Some of them are also named here: Micron, for example, Peloton, Wix, ClickUp and many more. We are so fortunate to work with them and build and shape this product together.
Kristin Myers: Now, you were saying something about before January that it existed as a different idea, I believe, and I had interrupted you. Tell us a little bit about that.
Yuval Zacharia: Absolutely. We started by working on a completely different use case. But when we worked, the three of us, myself and [unclear: “Aaron and Boris”], my co-founders, we built a proof of concept with Claude Code, which at some point just tried to open up our whole cloud environment to the internet. At that point in time, at that specific moment, we realized that if this is happening to us, you know, folks that are practicing security for, I don’t know, 50 years combined, it’s definitely going to happen or already happening to enterprises out there. And this is the moment when we realized that we need to start big.
Kristin Myers: So talk to us a little bit, for those of us that are not engineers and don’t have that cybersecurity background, what does that mean exactly, that you were being opened up to the cloud? Why is that so terrifying, and why is that such a problem for a lot of the companies that you do work with?
Yuval Zacharia: Absolutely. Look, enterprises are used to working in an environment where humans are responsible for the actions that they’re taking. Suddenly, what happens is that each and every agent that is acting on our behalf has the same access, the same permissions, like each and every one of us. This is when, you know, a DevOps engineer has access to the organization’s most private, most secure data, but suddenly an autonomous agent has the same access. It can do damage, right?
Kristin Myers: Okay. So talk to us a little bit about what the AI agent is actually doing inside of the company. You just hinted at it a little bit. They have access, obviously, to a lot of sensitive and private data, but how deep does it go that some of these AI agents are working? And what are some of the, I don’t want to say scariest cases, but I guess the most concerning cases that are really necessitating Bay to exist?
Yuval Zacharia: I would say that there are two different aspects to this same problem, right? One of them is an agent that can be compromised by an external attacker.
Kristin Myers: Right. This is attacked, hacked?
Yuval Zacharia: Yeah, absolutely. By a cybersecurity operator, or cyber operator, or offensive operator. Right. That is one aspect. But the other aspect is that agents sometimes can be just overachievers, right? They are trying really hard to accomplish whatever you’re asking them to do. And this is where intent drifting becomes key, in order to basically keep these agents in bounds and not let them go.
Kristin Myers: [unclear: “Confused.”] Does that happen often, that the agents get a little bit overzealous in their tasks and decide to go beyond the bounds that they’re supposed to work? Because everything that we keep hearing repeatedly about AI is that they’ve built these guardrails around what the agent can and cannot do. So how often is that drift happening?
Yuval Zacharia: So this drift actually happens a lot when we’re talking about the guardrails that the AI labs are providing. These are different guardrails in different steps. They can impact prompt injection and jailbreaking and a lot of other different scenarios that can basically happen to agents or that agents can cause. When we’re talking about normal infrastructure, enterprise infrastructure, sometimes critical infrastructure, the agent has power to do stuff that the AI lab cannot even know when training the model. Right? So when we’re talking about harnesses that have specific context in real time, but sometimes just trying to achieve a different purpose, this is where intent drifting can happen. And this is where you need Bay in order to put the right guardrails in place.
Kristin Myers: So there’s the drift. There’s also something that your team has written about. It’s called ghost jacking, and that’s where an agent reads a poisoned error log and then follows the attacker’s instructions. So there is a cyber attacker that puts these sorts of instructions inside of an error message, therefore poisoning it. So how does that happen? How does something that ordinary turn into a breach? That seems like something that we should have already, you know, solutions for.
Yuval Zacharia: Yeah. So the truth is that this world is moving so fast. There is a lot happening out there. And we’re not yet there in terms of the right guardrails and the right security. And this is a great example for that. So yeah, agents can read context and verify the input, from an error log or just, you know, over online or over Reddit or anything of that sort. This is impacting the next actions. And this is where we’re talking about intent drifting, and how we can put the agents in bounds and make sure that it doesn’t make any mistake, even if it doesn’t mean to. It’s not like it’s [unclear: “fold”] or anything, but no one is accountable for that. So this is where bounds are actually required, to steer the agents and make sure that it goes in the right direction, make sure that it’s obeying the organizational policy.
Kristin Myers: So talk to me a little bit about this tension that exists. Companies inherently want to go fast, right? Especially in the world of AI right now, being fast and moving quick on tech and on AI is huge. It’s a difference of millions and even billions of dollars for some of these companies. And so naturally they want to move quite quickly. Security companies and security folks are the ones slowing it all down. So how does that tension really get resolved between “let’s move very fast” and “let’s slow things down a little bit” in the name of cybersecurity?
Yuval Zacharia: Yeah. So I would say the opposite should be right. Because as a security practitioner, and working with a lot of security teams for years now, security teams don’t want to slow things down. They want to make sure that things are staying, you know, confident, safely. And their job is to accelerate the business and help the business grow. Specifically, we founded Bay in order to help them do that, to adopt AI as fast as possible, as much AI as possible, without breaking production, without leaking any sensitive information.
Kristin Myers: So if you want to know where Wall Street thinks security is headed, you just have to watch what the biggest players are buying. Palo Alto Networks, one of the largest publicly traded cybersecurity companies, has spent this year shopping for AI agent security. And then in February, it agreed to buy Israeli startup [unclear: “Koi”], the deal reported at around $400 million, all to keep watch over agents and extensions on employee devices. Then in June, it added Portkey, which sits between companies and their AI models to enforce the rules. The message from the incumbents: agents are the next big attack surface. And for a startup like Bay, that’s both validation but also competition. Now, Bay’s bet is that a platform built only for agents, deployed through tools that companies already run, can move faster than a giant stitching acquisitions together. But either way, the buyers are paying attention. We’re back now with Yuval Zacharia, co-founder and CEO of Bay. So, Yuval, let’s talk a little bit about some of the customers that you have. Again, Bay has been around for ten months, but you’ve moved fast. You already mentioned you work with several Fortune 100 companies, several that are listed here, to name some specifically, you’ve got Micron, Wix, ClickUp. You mentioned Peloton as well. What made them move so early with Bay, so fast after your founding?
Yuval Zacharia: I think, first off, you talked a little bit about the big guys out there. I think that something very important that a lot of customers understand, and specifically those that are working with us with Bay, is that the pace of AI, the pace the world is now moving at, requires a partner that actually can move at the same pace, and probably faster if possible. So this is the idea, and this is why these companies are adopting startups and technologies just like Bay. The customers that we are working with trust us to help them accelerate AI adoption, and again, not stop it, help them accelerate it. And as a result, in a few customers’ environments, we actually saw that AI agents can now run autonomously. This means without any human supervision whatsoever, three times more than it was without Bay.
Kristin Myers: So I’m curious to know, we just mentioned some of those clients that you guys have. Are there certain industries that you’re seeing right now gaining particular interest in a company like Bay and the security that it provides?
Yuval Zacharia: I think there are a lot of different verticals and sectors in the industry that are rushing to adopt AI. In some industries, most likely the tech industries, right, it’s about, right, so we adopted AI, now it’s the wild, wild West, and we need to understand how we are going to control these guys, right, that are running everywhere. On the other side of the house, the other sectors are a little bit more on the, okay, so we have to roll out AI in order to not lose the competition. How are we going to do that? How are we going to manage the deployment, run these guardrails, make sure that everything is in place?
Kristin Myers: So I want to know, because you were, or rather, I was actually going to talk about your background a little bit. But before I do that, so when a company does this, they add on Bay. You mentioned a second ago that they’re running three times faster, their AI agents are running three times faster with Bay than without. Are there other things that they’re able to do now, once they become a client and a customer of Bay’s, that they weren’t able to do before? Or is it really all about the fact that they can move more efficiently and with more speed?
Yuval Zacharia: With more, like, different harnesses, different tools, different AI tools, because we provide you the single control plane to control everything. So you don’t have to choose whether you want to go with Anthropic or OpenAI or Microsoft. You can just do all, and you can control everything with the same policies, with the same guardrails from a single control plane, which is Bay, but with the confidence and the security, knowing that your agents won’t go rogue.
Kristin Myers: So I guess the idea being here, that a company can feel more comfortable deploying those AI agents in more areas and more sectors of their business.
Yuval Zacharia: Absolutely.
Kristin Myers: Okay. So, stepping back a little bit, you were a major in Unit 8200, leading about 50 people, and you stepped off that track to train alongside 18-year-old recruits. Excuse me. So why start over? We love to find out a little bit about the histories of our founders. And so it was kind of a rare pivot to do, right? Nobody does that, really.
Yuval Zacharia: I just finished my military service and I said, hey, let’s do that all over again, right? And start as a, retraining as an offensive officer. I always had a thing for doing stuff that everyone tells me that I can’t. And I think that this is a big part of it. You know, when I joined this section of the unit as an offensive operator, there were five girls out of a department of 150 guys, and I was one of them. I was older than anyone else, and I just started to do this whole thing and learn everything from scratch and spend another four years there.
Kristin Myers: So, as I mentioned when we started after the break, Palo Alto Networks is buying agent security startups. When you hear something like that, that a giant like Palo Alto Networks is moving in on this space where you’re trying to currently build and scale, what do you think about that? How do you think about Bay competing against some of these giants? Because Palo Alto is obviously not the first and definitely not going to be the last.
Yuval Zacharia: Yeah, I think that Palo Alto is one example. A lot of others are buying a lot of AI startups. The difference is that, A, they will need to stitch all these different solutions together in order to actually come up with a value proposition that is sufficient and, you know, actually doing the job like Bay does. So that’s on one hand. On the other end, there is a full platform that can handle agents that are running on employees’ devices, over cloud environments, cloud workspaces, SaaS agents, basically follows the agents wherever they run, right? So this is the idea. As Palo Alto bought [unclear: “Koi”], for example, which is a point solution for the endpoint, they will need to buy probably 5 or 6 more companies in order to provide a full value prop like Bay, and then stitch them together, go to market with that. And we know, all of us know the pace that these guys are moving. It’s not fast enough.
Kristin Myers: I was going to say. So it sounds like you’re saying that even though these giants are bigger, have obviously deeper pockets right now and more clients, they’re way behind you in the race right now. Is that an accurate assessment, that they’re not even close to creating the platform that you currently have?
Yuval Zacharia: They are trying to start building it. Yes, I would say that.
Kristin Myers: Okay. So what has been the hardest part? You’re ten months in, I’m doing the math right, where we started in the first month, we’re now in the 11th, ten months in. What has been the hardest part about building Bay so far?
Yuval Zacharia: Wow. There were so many.
Kristin Myers: Talk to us a little bit about that, some of the hurdles that you guys have experienced.
Yuval Zacharia: I think that every day you experience so many, you know, dopamine and then so many struggles and so many hard things that are happening. Sometimes it’s an engineer that says, no, I went to another company. Sometimes it’s a customer that says that he chose to go with a different competitor. And then the right moment afterwards, something just happens, or you’re signing a new logo, or you have a conversation that gives you an idea for, you know, the entire, a few milestones for the roadmap, or you’re getting some signals from the market about what you’re doing, why it’s important and how they use it. So it’s all ups and downs.
Kristin Myers: As you think about the next year, three years, five years with Bay, and you think about expansion and you think about scale, are you guys focused right now on acquisition of more clients and more customers, or is there something that you’re also hoping to do with the platform as well?
Yuval Zacharia: Yeah. So absolutely double down on the platform. We’re building a true platform. So multiple products, again, following the agents wherever they run. Doesn’t matter if it’s on the laptops, over the cloud, on gateways, on SaaS, everywhere.
Kristin Myers: So are there products that you don’t currently have yet that you’re hoping to?
Yuval Zacharia: So we are building it. We’re building it in phases. Most of it is in place, and we’re expanding also our breadth and depth of each and every solution that we currently offer to the market, continue building with our partners, with our customers, in order to make sure that everything is on spot, providing the right answers, and of course the pace that things are moving. Every other day there is a new thing that you need to support, new thing that is happening. You have to keep up. And this is exactly what we’re doing. So building the platform, absolutely. The other side is of course go to market, establish our brand. We’re launching the company today. So it’s a very exciting milestone. We have to build this recognition, to be out there, of course working with more customers. And that’s exactly what we’re going to do.
Kristin Myers: I have to ask, as the last sort of question, as you think about, you know, we’re here on the floor of the stock exchange, do you think about going public? And when do you think that will happen? Or is this something that you think perhaps one of those larger companies will come along and snatch you guys up, instead of trying to stitch together those platforms, they say, hey, we could just buy Bay, it’s a one-stop shop.
Yuval Zacharia: Absolutely. This is the dream, okay? This is the dream. This is what we want to do. We’re going to go all in. Going to go big. Look, there are not a lot of opportunities in life to be that kind of company. Work is changing completely. AI is changing everything that we knew, and enterprises are operating completely differently. And this is exactly the point in time where a company like Bay has this kind of opportunity to change the way enterprises work, end to end.
Kristin Myers: Well, we’re so glad that you joined us on your launch day. And if you’ve already gotten this far in ten months, I imagine that ringing that bell is going to happen sooner than later. Yuval Zacharia, co-founder and CEO of Bay. Thank you so much for joining us.
