Hello and welcome to fabulous Las Vegas, Nevada. We're here with a special exclusive series on theCUBE talking about resilience and AI and cybersecurity. Three of the most important topics, not just of this year, but of the next five years of the decade. My name's Savannah Peterson. Very excited to be joined by one of the brightest voices in the space. Bipul, thank you for coming to hang out.
Savannah, such a pleasure.
This is a joy. You're a CUBE OG. You've been a regular on the show. I feel so lucky that I get my turn with you. This is going to be awesome. AI definitely has been the acronym of the year. Talk to us a little bit about what Rubrik's doing in AI right now and why you're here in Las Vegas.
As you know, Rubrik is the security and AI operations company. We deliver cyber resilience to ensure that our businesses, customers are always in business and their services are always up and running, even when they are faced with cyber attack. And then we have launched a new product line recently called Rubrik Agent Cloud. And Rubrik Agent Cloud is all about understanding what agents are running in your company, what are they actually doing? What are the guardrails around them? And if they misbehave, how do you undo those agents? So, essentially, Agent Monitor, Agent Govern, and Agent Rewind is a complete Rubrik Agent Cloud solutions that we are selling. So we want to ensure that our customers can do AI transformation with confidence.
That seems to be one of the big themes. I'm glad you brought up confidence, uncertainty. Definitely a theme right now as people start to scale up, whether that's at the enterprise or even in the small business scope. Let's talk about how you've been navigating, making sure that you delivered the right product. I mean, you're dealing with incredibly sensitive data. And to your exact point, all of these agents create a bunch of data that also needs to be governed and compliant. When did you start building this? And talk to me about the process to make sure you were able to keep up. Everything's moving so fast.
As folks have gone into the agentic work, what they're realizing is that these agents can do 10x more damage in 1/10th of the time.
I'm glad somebody said it.
Because they're running business processes. They are assuming your identity. In some ways they are super human. And if they get compromised by a threat actor in North Korea, now that threat actor is running your business operations, which is very distinct from stealing your data. And so for businesses to have the confidence and take advantage of the 100x opportunities that AI provides, they need to really contain that risk. And this is what we have been focused on. We said taking AI agent is like hiring 1000 employees and letting them loose to write your code, to serve your customers, to approve your purchase order.
It's like a Black Friday rush all at once is what I'm thinking about.
Exactly. But how do you do that without creating governance around these actions, around these activities? And if they misbehave because of hallucination or otherwise, how do you ensure that you take out the unwanted actions?
Yeah. I mean, first of all, I actually would love your take on this. Where do you think most companies are in terms of their implementation and adoption journey when it comes to AI? Is everyone still just doing MVPs and figuring it out in the sandbox? Or are you helping them ensure their security at scale already?
Every enterprise has started with doing the basic AI, which is around search and answers. And as they moved into the agentic world, they saw good results in their pilot. But going from pilot to production, they have to really take care of the operational risk. And that is what is the biggest inhibitor because the issue is that in production, you need to have accurate AI. You need to have cost control. Otherwise, if you're doing a lot of inferences, your cost will run away. So, accurate results, cost control, creating governance, and making sure that they are doing the right action is not an easy thing to achieve.
That's why every IT leader and security leader is looking for solutions around compliance and governance.
Absolutely. And it's super complex. I mean, when we're talking about multiple nations housing data, when we're talking about who gets to see that data. So, talk to me about how you prioritized which solutions to provide to your community first?
It is about layered cakes. The base layer is, do people understand what the hell is going on? And in terms of what the hell is going on is about how many agents do I have? These agents could be, you develop these agents, we call it infrastructure as a service agent. Or you could take an agent from a platform such as Salesforce or M365. So do you have an inventory of all your agents? And do you understand what these agents are doing? That is the base layer. Once you do that, then you say, 'What are my broad governance?' It's like hiring an employee. When you hire an employee, you give the employee a handbook saying what is allowed and what is not allowed in the organization. Do you have that level of governance as a base governance across all the agents? And then just like an employee that works in a specific department to do a specific job and what is allowed and not allowed for that employee, how do you do that for agents? How do you ensure that agents, when they assume an identity and work on the data, they can do certain things and cannot do certain things. So, creating guardrails around these agents, making sure that agents are producing accurate results at a contained or lower cost is a real challenge.
It's a huge challenge. And it's fun to get to chat with you because you see the billboards, 'We have 50,000 agents doing this.' Or, '1 billion agents by X date,' or whatever that might be. And I look at that and think that's one billion points of liability and one billion opportunities for risk. So, let's talk a little bit about where you think agentic is going to derive the most value in general. You get to see a lot of different applications. What is this going to change in our world?
If you look at businesses, what they're seeing is that agentic work can really create an enhancement to their line of business, whether it is sales or marketing or customer service or writing code or just business operations, it can make a difference. So there is a huge pressure from the board and CEO to do this transformation. And it is one of the first technologies after a long time where the pull from enterprise is unparalleled.
Everybody's now seeing that, how do I transform my business? And it is transforming the job of CIOs. CIO's job used to be consume the services from cloud or otherwise and create a portfolio of services to reduce risk. Now they are actually doing knowledge work for line of businesses while reducing risk. It has taken their job one order of magnitude more complexity and now they're thinking about how do I create this agentic AI enterprise without undue risk?
What would be your one piece of advice to someone trying to navigate that right now outside of obviously deploy Rubrik?
One piece of advice is that figure out the workflows that you want to automate because there's no such thing as AI transformation. Figure out specific workflows. And start with three or four specific workflows that are manual or time-consuming and in many cases can't be done without AI and implement AI and prove them. Once you are able to do that, you will learn how this whole process works, what are the guardrails you have to create, how do you train the organization? Because there's a lot of people and cultural change has to happen to adopt the agentic work.
So, ability to deliver three or four such workflows in a fully agentic model is the baseline. After that you scale rapidly.
