ITTech Pulse Exclusive Interview with Devvret Rishi, GM of AI, Rubrik

Devvret Rishi is GM of AI at Rubrik, chats about focusing on governed autonomous agents, cyber resilience, and safe enterprise AI adoption in this catch-up with ITTech Pulse.

How are you inspired by and approaching AI?

The primary thing that inspires me is seeing how some of the leading technology companies and “SF-based AI startups” have started to use AI in their everyday workflows. The amount of leverage they get out of the tools is a night-and-day difference when speaking to a traditional vs AI-native organization. This momentum inspires me to accelerate a higher level of AI adoption for the enterprise, starting to move toward systems that use AI for more than just better search and information retrieval – and have systems that are able to actually perform tasks.

Right now, most AI is reactive, it waits for a prompt. At Rubrik, we’re focused on the next horizon: agents that manage end-to-end workflows. Imagine a DevOps “Watchdog” agent. It detects a sudden server spike, digs through the logs to find the error, identifies a memory leak in a recent PR, and then pings an engineer with a one-click automated rollback. That’s the level of efficiency the industry is building towards.

Read More: ITTech Pulse Exclusive Interview with Linir Zamir, Team Lead-AI Engineer at Telit Cinterion

My approach is rooted in practicality, not just hype. When I joined Rubrik, I sat down with 180 executives to understand what was actually keeping them up at night. The consensus was clear: the bottleneck is no longer deploying AI models or building agents, its managing risk associated with autonomous systems.

Most organizations are stuck in “read-only” mode because they are rightfully terrified of what happens when an agent takes an unscripted action in a production environment. To move the needle, we’re building governance systems that provide the guardrails necessary to evolve from simple prompts to autonomous agents. We’re bridging that gap by focusing on observability and real-time remediation, giving teams the peace of mind to let the AI actually go to work.

What’s the one moment at Rubrik that made you think, “We’re building something special here”?

The experience of joining Rubrik following the acquisition of my startup, Predibase, and the subsequent four months spent meeting with 180 executives stands out. Seeing the reaction of enterprise customers when they heard the message that we were thinking about managing risk with AI – that really started to inspire my confidence. And then gaining those insights allowed me to define a clear four-phase path for agentic AI adoption. However, the real lightbulb moment came when I recognized that Rubrik has a superpower no other organization has.

Since we already sit at the intersection of data, identity, and application context, we are the only ones who can offer a true “undo button” for AI. Witnessing IT and security leaders realize they can rewind an agent’s mistake, the same way they’ve trusted Rubrik to recover their data from disaster and cyber threats, was the moment I knew we were building something special. We are giving organizations the confidence to transform into AI enterprises.

Everyone talks about partnerships – what makes your AWS deal different from the usual vendor collaboration?

Our collaboration with AWS at re:Invent addresses two distinct but connected problems at once. First, we announced the integration of Rubrik Agent Cloud with Amazon Bedrock AgentCore. Now users building agents on Bedrock, as well as any agent builder tool, can easily enable Rubrik’s monitoring, governance, and remediation tools for agents to drive trusted AI adoption.

Read More: ITTech Pulse Exclusive Interview with Oleg Sapon VP of Product Management at MSP360

Second, we achieved the AWS Resilience Software Competency as a launch partner. The recognition validates that Rubrik’s foundational platform provides the design, recovery, and operations that customers need for their most critical systems. That fundamental level of trust is absolutely essential – you can’t safely build autonomous AI systems without a reliable recovery posture underneath them.

Cyber resilience meets AI safety in this partnership. Why now? Why is this combination such a game-changer?

We recognize that the basic, foundational trust in data recovery is essential for high-stakes AI operations. Simply put, organizations need a strong, proven recovery plan for their data before they can deploy independent and secure AI systems. Now is the time to prioritize this because the risk is accelerating rapidly: AI agents are multiplying in the workplace, acting as new, highly autonomous identities with wide-ranging privileges. If an agent makes one small mistake, it can instantly compound into a disaster – like deleting an entire database. By combining our proven cyber resilience with the specific AI safety controls of Rubrik Agent Cloud, we are helping customers solve the challenges at this critical intersection of security and AI acceleration.

You’re asking enterprises to govern AI at scale with Rubrik Agent Cloud – what’s the hard truth they need to hear?

The hard truth is that AI will likely increase productivity by 10x over the next several years, but it will also come with significantly greater risk. Developers are rushing to install AI agents without IT oversight or a proper security review, which leads to organizations struggling with the basic visibility and governance of these agents. As a result, IT teams generally lack a single pane of glass that gives them a view into which agents exist, what data they access, and what actions they take, creating a massive governance blind spot. To help address the liability challenges that come with high autonomy, they need a tool that maps their agent landscape, provides custom guardrails, and explains the root cause of an agent’s mistake with the ability to block or reverse high-risk agent actions.

It’s late 2025 – where’s the biggest opportunity for Rubrik and the whole data resilience space?

Our biggest opportunity in 2026 is to double down on foundational security trust and accelerate safe AI adoption. We plan to continue our focus on AI agent security, with further development of the Rubrik Agent Cloud platform to push AI adoption forward while providing clear, accountable guardrails against risks. This really comes down to making sure that as companies lean into AI, they aren’t trading off their core security to do it.

We are also focused on helping organizations modernize their current IT and security workforce management, as robust security and governance systems will be essential to monitor, control, and remediate agent output. This is especially important given that the proliferation of AI agents is creating the “great AI sprawl,” forcing IT and security teams to reconcile rapid deployment with system control. As AI agents start handling more tasks, the goal is to provide the visibility and tools needed to manage that output effectively, ensuring that these new automated workflows remain safe and reliable for the long term.

One piece of advice to leaders: how do you actually balance moving fast and staying secure?

We can’t balance moving fast and staying secure by just using people, paper, and processes. We need technology solutions to help enable secure adoption of AI. Those technology solutions are going to start with observability, which gives you a strong lens into what exactly AI agents are going to do – because you can’t manage and govern what you can’t see. But the best solutions will go beyond that and actually give you power you can use to enforce your requirements on AI systems at run-time.

Today, many organizations think of solving for AI risk procedurally – with committee review meetings. That is going to be too slow for the modern enterprise. Instead, you need to develop a framework, encode it into a technology platform, and allow that platform to run your framework in real-time for you.

Thank you, Mr. Devvret, for taking the time to share your insights with us.

Write to us [⁠wasim.a@demandmediaagency.com] to learn more about our exclusive editorial packages and programmes.