theCUBE + NYSE Wired: Zero Trust Cyber Series | Nishant Doshi, Cyber Haven

Clips
More from theCUBE + NYSE Wired: Zero Trust Cyber Series

Nishant Doshi

Chief Product Officer

Cyberhaven

play_circle_outline Safeguarding Sensitive Data: Cyberhaven's Innovative Solution and Data Lineage for Secure Information Management

play_circle_outline AI engine for anomaly detection and predicting user behavior

play_circle_outline Lightweight endpoint agent in user mode for minimal disruption and efficient data collection

play_circle_outline Innovative Companies Embrace Data Lineage, AI Security, and Market Expansion for Anomaly Detection

Info
Transcript

Nishant Doshi, Cyber Haven

Nishant Doshi

Chief Product Officer Cyberhaven

John Furrier welcomes Nishant Doshi, Chief Product Officer of Cyberhaven, to discuss their data security solutions. Cyberhaven focuses on protecting sensitive data from being exfiltrated outside the organization, particularly in the face of increasing cyber threats like ransomware and insider threats. Their platform uses a lightweight endpoint agent and a purpose-built graph database to track data lineage and detect anomalous activity. This approach allows them to provide context and accuracy in data security, which traditional solutions have struggled with. ... Read more

explore Keep Exploring

What does Cyberhaven do to protect sensitive data for innovative companies? add

What new solution is being implemented in data security to increase accuracy and address the issue of anomalies detection in data exfiltration flows? add

What is the difference between user mode and kernel mode in a system, and how can sitting in user mode be advantageous for certain companies in terms of performance and security? add

What is the key innovation that sets our data security solution apart from others in the market? add

bolt Powered by CUBE AI

Nishant Doshi, Cyber Haven

search

>> Hey, well, hello everyone. Welcome to theCUBE here at NYSC. I'm John Furrier, host of theCUBE. This is our East coast studio on the show floor. big day today. A lot of activity going on. This is part of our media week. We're here with three days of wall-to-wall coverage on Wall Street, cybersecurity, AI innovators, all coming together. Nishant Doshi is here, the Chief Product Officer of Cyberhaven. Nishant. Great to have you on. theCUBE. Great to see you. Cyber is our favorite topic this week. Obviously one of the key themes this week in town. A lot of cyber events. Obviously there's an AI event going on. Thanks for coming on.

Nishant Doshi

>> Oh, thank you for having us.

>> Not too bad here.

Nishant Doshi

>> Not too bad.

>> This view behind us.

Nishant Doshi

>> It's an amazing view. Yeah.

>> I love the NYSC. It's like a Hollywood set. It's beautiful. Innovation's here. Capitalism is here. We're in Palo Alto with SiliconANGLE. You're from the Bay Area. We live in that world connecting together. Cybersecurity is a data problem. We've talked about that. You guys are at the center of it. Data security's huge. I want to get into data security, what you've seen with customers and some of the challenges. Before we get there, talk about Cyberhaven, what you guys do. We'll just set the context and then we'll get into it.

Nishant Doshi

>> Absolutely. Yeah. So Cyberhaven, it is a data security company. We are a CDC company and we protect the most sensitive data for the most innovative companies. Our solution is a data detection and response platform, and you can think of us as a next generation DLP solution, a data loss prevention. So what we do simply put is we protect data sensitive, data from being exfiltrated outside the organization.

>> Which everyone's trying to protect because it's happening a lot.

Nishant Doshi

>> It's happening a lot. So especially with GenAI, with a lot of nation-state attackers, that sensitive data is very prone to sort of-

>> And it's great for ransomware too. That's the core negotiating leverage on all the ransomware action.

Nishant Doshi

>> There's a lot of threat vectors and ransomware's definitely one of them and-

>> And what are some of the other vectors?

Nishant Doshi

>> So broadly speaking, there's two buckets of data leaving the organization. One is malicious and the other one is inadvertent. On the malicious side, you just have a lot of what I call human supply chain issues. You have employees who are leaving the organization, they might take some data. There's some very famous cases, a lot of IP gets leaked that way. Then you have these nation-state attackers. You have employees who are motivated and planted there to kind of take-

>> Like insider threat?

Nishant Doshi

>> Like insider threat. Exactly.

>> That's what they call insider threat.

Nishant Doshi

>> That's the insider risk and insider threat. Exactly. So that's a big problem for a lot of companies. And in this day and age, IP is everything. So when you have all of the sensitive IP and you have an insider who's motivated, who has access to this information, you really need to figure out how to protect your data. And that's where Cyberhaven really shines.

>> And there's no debate that the market trends are there for protecting more than... The scales used to be. Chipped so bad in favor of the bad guys, now we're starting to see GenAI come up a little bit. So that's good news. So there's no need to talk about the need business model. Talk about your business, your business model. How big are you guys? What are your deployment options? What's the consumption look like? How does the customer engage? Talk about the business.

Nishant Doshi

>> Yeah, absolutely. Yeah, so we just raised our CDC earlier this year. We are very well capitalized. We just raised about $88 million earlier this year. Our solution, I'll talk about our solution first. So our solution is essentially an endpoint agent that sits on your endpoint. It's a very lightweight agent. That's one of our key innovations. That our agent is very lightweight, has visibility into every user interaction, every sort of activity that the user's doing, the employee's doing. Whether it's opening a file or sharing something or copy-pasting some document. So we have access to that. Our second innovation is really taking all this information and stitching that graph of how data is moving through the organization. We call that data lineage and the idea of data lineage is where data starts and the journey of that data through the organization. So I could be downloading a file from Salesforce, an application, sending it to five people, they could send it to another five employees and then one of them could leak out some information. What we are doing is tracing the entire lifecycle of that data, and by doing so we do two things. We can understand that data really well and that's the holy grail of security or data security.

>> It gives you context too.

Nishant Doshi

>> It gives you context exactly. So a lot of traditional approaches to data security for the longest time have been focused on understanding the content using regular expressions and pattern matching technologies. That has not scaled well. So the innovation that Cyberhaven sort of brought to the table was to use context and the context that we use is data lineage. When we use data lineage, it's the old saying, where you know where you're coming from, you know where you're going. So we understand where the data's coming from. So if we know the data's coming from certain applications, certain users, we know where the data has been accessed, we have a better sense of what that data is. Of course we do content inspection as well. So together, it's a very powerful solution and it increases the accuracy, which is a problem that's plaguing this data security work for the longest time. So we are really disrupting this-

>> So you're talking about anomalies detection too. If you know certain behavior, you know when you can recognize bad behavior, and that's been hard to do you're saying?

Nishant Doshi

>> That's been really hard to do. So now we have these billions of flows that we've collected for every customer and for those customers what we do is we put a new engine, which is an AI engine. It's called Linear that we are just launching it soon. But the idea is to use that data flows to find anomalous activity and anomalous data exfiltration flows. So we can predict what a user is going to do next because we have that history and we are using-

>> So you're setting up beautifully for GenAI because you're one, doing the foundational, grinding it out and doing the work, the lay-out, the data sets, and you're data driven heavily because there's a lot of data involved.

Nishant Doshi

>> Absolutely, yeah. We understand the data and the context of that data really, really well, which sets us up for success in this new world. The other thing is our IP is really the agent, which is very lightweight. The fact that we can sort of have an enterprise wide lineage, which allows us to connect the dots really easy.

>> Talk about the agent for a second. Lightweight, what does that mean, lightweight compared to what? What's heavy? What's light and where does it sit?

Nishant Doshi

>> Yeah, so that's a great question. So traditionally agents have been in the kernel mode. So in a system you have a user mode, kernel mode. A lot of agents, traditional companies, incumbents, they have a lot of kernel mode agents, and then you have outages and we saw one a few months back.

>> So it's potentially disruptive or is it overhead from a performance standpoint or both?

Nishant Doshi

>> Both. Both. You could actually have essentially take down the entire systems, so I-

>> We saw that with CrowdStrike and with Microsoft. That was kernel level.

Nishant Doshi

>> We have seen that with the vendors for sure. And I would say that for us, we sit in the user mode. So we sit in the application layer and we still get access to all the information that we need, but by virtue of sitting in the user mode, we are pretty lightweight and we can understand... We get a lot of the information, and then our cloud is really the brain behind this lineage and understanding and connecting the dots. So the deployment model is that you deploy these agents and then use the Cyberhaven cloud to manage-

>> Manage, orchestrate-

Nishant Doshi

>> Orchestrate.

>> Analytics.

Nishant Doshi

>> Analyze, policies, alert. Essentially, produce this.

>> That's your power source. The cloud's your-

Nishant Doshi

>> Cloud's the power source.

>> That's your supercomputer that does all that, the work, analytics. Give me the reports, do some work-

Nishant Doshi

>> Exactly, exactly. Yeah, the telemetry goes to the cloud and then the cloud is really where you do all-

>> I really like how you laid out the agent lightweight versus heavyweight with user mode and kernel mode because that really gives you a perspective of where you sit in the stack. I'm sure customers must like that.

Nishant Doshi

>> Customers like that, and it's widely deployed in the most innovative companies.

>> You're kind of getting out of the way, let the kernel do its thing. You're getting everything you need, you get the cloud behind you to power it.

Nishant Doshi

>> Yeah. So that's one source. And then I would say the key innovation is the data lineage. And for that we have a purpose-built graph database, which is allowing us to connect these data flows in a very fast way. So if you take a commercial graph database, you can only do a few hops and after that it slows down. Our purpose-built graph database allows us to connect these dots.

>> You can traverse fast.

Nishant Doshi

>> And traverse free in milliseconds.

>> How deep can you go on the graph?

Nishant Doshi

>> Yeah, we can do 200 hops in milliseconds.

>> Milliseconds?

Nishant Doshi

>> Yeah. And usually for a commercial graph database, seven hops and then it slows down quite a bit.

>> That's great.

Nishant Doshi

>> It goes into minutes and-

>> Yeah, it's funny, Nishant, how graph databases are hot now. I've always loved graph databases. I remember-

Nishant Doshi

>> I'm a technologist and I love graphs. I've always had a fascination with graph databases.

>> As why is graph databases important right now in your opinion? Because I think this is... I mean, I know why I like them, but there are nodes in ours, which is great for computer science and code. You can parse, traverse, and get data fast.

Nishant Doshi

>> I think it's all about context. Again, so when you're looking at data, especially because that's been the hardest problem, infrastructure, application security, there's a lot of companies at a hundred million dollars or billion dollars plus in this market. But data has been the unsolved problem. And the reason data is hard is because you need a lot of context. My data is going to be different than your data. When you think about infrastructure, my port 80 is the same as your port 80. It means the same thing, but it's not the same thing with data. So in order to do data analysis, you need a lot of context and that becomes a multivariate problem. It's not one dimension. And because of that, it yields itself to something like a graph engine because you need to connect the dots across many different things.

>> Yeah, general purpose stuff is not winning in the new world. By the way, on the infrastructure side, it's getting customized too. If you look at all the top clusters on how the distributed computing paradigms evolving quickly, it's cloud connected from the enterprise to the premise to the cloud and edge. That's distributed computing basically, but that also might have different requirements based upon the deployment options. So that's new data too.

Nishant Doshi

>> Yeah, absolutely. Right. So you have data in all these different sources. You have the endpoint data, you have the PaaS data, SaaS data, IaaS data, these all different sources of data, and ultimately you need to understand. Now in that problem, the unstructured data, the human-generated data, that's the hardest problem because again, you can double-click on that. That's context again. My context is different than the document you're going to produce and then that sort of has a-

>> Okay, so I have to ask you, what's the secret sauce on managing that context challenge? Because I think that's the holy grail because you've got the behavior down with the lineage. Context is critical. Contextual behavioral data sets are used to manage context.

Nishant Doshi

>> So we use the lineage for that context. So to give you an example, if someone was exfiltrating of a sales roster to their personal Gmail account, traditional solutions, you would have to look at the content and pattern match and say, "Hey, this is sales data." What we can do is since we know that that data is coming from Salesforce in some shape or form, whether you've copied it, you've copied two slides out of that, doesn't matter. Through the lineage-

>> That's sales exporting all of the context.

Nishant Doshi

>> Yes, exactly. But we know the lineage. They can encrypt it, they can do whatever they want, but we know where it's coming from.

>> So that's the secret.

Nishant Doshi

>> That's the secret. So now we know without even looking at the content, Hey, where are you coming from, and that's-

>> So back to your point about not only hops on the graph, getting that at milliseconds, the content inspection overhead is gone if they use tactics to hide it.

Nishant Doshi

>> We have a better together story. So we do content inspection, but we also use lineage to classify data and that together is a very powerful solution.

>> So do you have a lot of reasoning going on? Because I've been on this whole reasoning bandwagon for over a decade I've seen, and now it's hot. Meta reasoning. Metadata is... I mean, re:Invent, one of the things I walked away from re:Invent was S3 tables that they announced on the storage, which is going to revolutionize the data, is that reasoning engines are becoming more powerful.

Nishant Doshi

>> We have a reasoning engine which is built on top of our lineage, which is called Linear. And what it does is really interesting. So we've sort of innovated on a technology that we call large lineage model.

>> I like it.

Nishant Doshi

>> Not large language models. And the idea is we are predicting the next action. We are also looking at the content and we're sort of reasoning on doing reasoning on the content itself. So a great use case for us is when sending screenshots out, and that's a very common data exfiltration pattern that people are sending these screenshots because they take a whiteboard screenshot or a picture and then send that out. And it was very hard to operate or understand that data at scale. With AI and with the reasoning engine, we can discern between a cat meme versus a schematic. And not only that, we can actually give a lot of color and a lot of context on where that schematic was from, what it's saying about the company or the IP. So we can use that information to do a much better job. And that's really the key innovation there.

>> I really love what you do, and you're the Chief Product Officer. By the way, I think you're set up for GenAI beautifully. By the way, metadata is a big topic that was discussed at re:Invent just last week. And again, not surprising to us. We've been following this for a long time, but the importance of the metadata, lineage data, that's just... I mean, I consider that metadata, but you probably disagree, but I'm oversimplifying it. But it's not about that one piece. It's the data around the data.

Nishant Doshi

>> Exactly. And that's the key. It is metadata to your point, and it's about not only that data, but who's accessed it-

>> Which is metadata too.

Nishant Doshi

>> The folder, where the source is, all of this is metadata. All the metadata, the more metadata you have, the more context you have.

>> As they say, it's meta of meta. Nishant, you're the Chief Product Officer, you got the keys to the kingdom. You have to manage both the customer requirements, which some would say is a moving train in this market. It's super high velocity. Pace of play is very high. And then also manage the engineering management side of it, product and engineering, which by the way, we're in a highly agile accelerated play there too. So you're in the hot seat. How do you do it? How do you deal... I mean, it's changed a lot. What's changed on your job? Go back a decade or two, it's do the PRD and we're done, send to engineering, QA the product. Now it's like push code every day. Take us through a day of what it's like to be a product leader in this market.

Nishant Doshi

>> So the role I have is becoming more common now, which is product and engineering coming together. So I really believe in that because it can make quicker decisions. Again, going back to the new way of doing things, which is you want to minimize the decision loops and that's really sort of where we can maximize our speed. The other thing is yes, I would say it's again people process and of course the product. It's having a great team and a great leadership team that you can rely on. It's having really scalable processes, but these processes are for geared towards urgency and speed.

>> Speed, yeah, speed. Fast. Yeah, move fast.

Nishant Doshi

>> Speed is everything.

>> The key is getting buy-in from the team.

Nishant Doshi

>> Getting buy-in.

>> I hate to use the term buy-in. Steve Jobs has that famous line. "If you need buy-in, well-"

Nishant Doshi

>> "You don't have the right-"

>> "The right team."

Nishant Doshi

>> Exactly.

>> They're either bought in or not. Jeff Bezos has the same philosophy that was handed down to Andy Jassy, which is debate, align and commit.

Nishant Doshi

>> Yeah, it's a very common pattern. So we have a disagree and common principle, which is very similar. We move fast.

>> Nishant, thanks for coming on theCUBE. I'll give you the last 30 seconds. Put a plug in for the company, value proposition. You're going to hire what areas, what's your growth strategy? Give a quick plug, take a minute.

Nishant Doshi

>> Yeah, absolutely. Cyberhaven, we are a data detection and response platform. We really are the leading vendor when it comes to data security and when you want to protect your data from exfiltrating as well as understanding your data and the flow of your data in your organization. Our innovation is really understanding not only the data but how that data is moving through your organization. And we do that using a key innovation, which is data lineage. We are one of the only vendors who could do this data lineage at scale for enterprise customers. And using this lineage we can understand that data, classify that data and protect that data with a degree of accuracy that no other vendor can. So that's really a key innovation. We're not stopping there. We've added our AI engine on top of that, which is amazing. It allows us to resume the data. It allows us to find anomalous activity and anomalous data patterns, exfiltration patterns. So that's something that's also really, really a key because you can't manage this by policies alone. So we are taking and attacking a domain that has been really hard to operationalize. It has been really hard to sort of manage. And we have a very new approach to data security. And luckily for us, we are seeing the most innovative companies in the world adopt these solutions at scale and that's really given us a very good position to kind of expand to a lot of different vectors in the data security market as well as take advantage of some of the AI technology and then of course protect AI as well in the future.

>> All right, Nishant, thank you for coming on the team. Good to see you again.

Nishant Doshi

>> Thanks a lot.

>> Thanks for coming on again, Nishant. For theCUBE, we are here at the NYSC. This is part of our East Coast access points, our super node, our super studio. We got the Palo Alto and NYSC Studios, which will become the backbone of the Wired Network. Put on by the NYSC, Brian Bauman and theCUBE team. I want to thank everyone for watching. See you next time.

Nishant Doshi

>> Thanks a lot.