NetApp INSIGHT 2026 | Sandeep Singh, NetApp & Helen Yu, Tigon Advisory Group

Clips
More from NetApp INSIGHT 2026

Sandeep Singh

SVP & GM, NetApp Platform

NetApp

Helen Yu

Founder & CEO

Tigon Advisory Group

play_circle_outline Unified Data Infrastructure and Operations for Distributed Estates: Novus AI Factory 100 TB/s Zettabyte Namespace

play_circle_outline Human-Agent Autonomous Control Plane with NetApp Console for Intent-Based Optimization and Policy Enforcement

play_circle_outline Trustworthy Autonomy: Transparency, Guardrails, Decision Logs and Audit Trails, Human-in-the-Loop Boundaries, Consistency, Decision Quality, Adaptability

play_circle_outline Measuring automation success: outcomes, metrics, governance, RACI including agents

play_circle_outline NetApp Insight 2026 Keynote: Metadata-Driven Discovery, AI-Ready Data and Zero-Copy Governed Access

Info
Transcript

Sandeep Singh, NetApp & Helen Yu, Tigon Advisory Group

Sandeep Singh

SVP & GM, NetApp Platform NetApp

Helen Yu

Founder & CEO Tigon Advisory Group

In this interview from NetApp INSIGHT 2026, Sandeep Singh, senior vice president and general manager of enterprise storage at NetApp, joins Helen Yu, founder and chief executive officer of Tigon Advisory Group, to talk with theCUBE's Rebecca Knight and theCUBE Research's Christophe Bertrand about how unified storage and an autonomous control plane are becoming the foundation for AI-ready enterprise infrastructure. Singh details how a consistent data infrastructure strategy bridges on-prem, public cloud and neocloud environments, highlighting the newly announc... Read more

explore Keep Exploring

Do the recent announcements and technologies risk creating a new data silo? add

How does NetApp Console's autonomous control plane and agent-based approach enable intent-based outcomes and continuous optimization for customers with distributed data estates? add

How can trust in autonomous systems be built and assessed, including the role of human‑in‑the‑loop, policy boundaries/guardrails, and metrics such as consistency, decision quality, and adaptability? add

What should leaders look at when assessing whether a process has genuinely improved? add

How is NetApp enabling customers to prepare for and lead in the AI era through its data infrastructure, governance, and access strategy? add

bolt Powered by CUBE AI

Sandeep Singh, NetApp & Helen Yu, Tigon Advisory Group

search

Rebecca Knight

>> (INTRO)Hello everyone. Good afternoon and welcome back to theCUBE's live coverage of NetApp Insight 2026. I'm your host, Rebecca Knight. I'm sitting alongside Christophe Bertrand, who is theCUBE research analyst in residence. And we've got two terrific guests for this next segment. I would like to first welcome Helen Yu. She is the founder and CEO of Tigon Advisory Group and also host of the very popular CXO Spice podcast. Welcome, Helen.

Helen Yu

>> Thank you for having me.

Rebecca Knight

>> Yeah, first time on the show. We're delighted. And we have Sandeep Singh, of course, returning to theCUBE, a CUBE veteran, I should say, SVP and GM NetApp platform here.

Sandeep Singh

>> Hi, Rebecca. Hi. Great to be here again.

Rebecca Knight

>> Welcome back. So I want to start with you, Helen. Tell us, tell our viewers a little bit about Tigon Advisory Group. You started in 2017, an idea before its time and a really good one.

Helen Yu

>> Yeah, we started off with becoming a fractional CXO service to startups because for the upcoming companies, they can't afford having a full executive team. So they're paying us the price of one to get 1 day of CRO, 1 day of CMO, 3 days of Chief Product Officer. And that morphed into AI deployment back in the day, right? We were building the chatbot for a global bank at the time before everyone even talked about AI. And then I'm doing what I'm doing. I never thought I'd be doing today sitting here with you because I was more of a builder and creator. So I'm so delighted to be here with you today. Yeah.

Christophe Bertrand

>> So, Sandeep, I'm going to start with a question about silos because data is everywhere today. There are multiple silos. It's in the cloud, in the multi-cloud, in the hybrid cloud, in the Neocloud, or on-prem actually, maybe even more so, which is funny. I like it personally. Are we creating a new silo potentially with some of the recent announcements and some of the technology you're working on?

Sandeep Singh

>> Absolutely not, is the short answer. Christophe, as you mentioned, data is everywhere. Customers should have the absolute freedom to be able to deploy their application workloads where it makes the best sense for them, whether it's on-prem, whether it's in the public cloud, whether it's Neoclouds. We know that the customers are going to use a variety of locations for the right fit for them. What becomes fundamentally important for customers is to have a consistent data infrastructure strategy that underpins it all so that they can get a consistent set of data capabilities combined with a consistent operational experience across the various environments. And when you mention the announcements, It's one underlying data infrastructure architecture manifested in different ways for customers. So whether it's the highest performing, the largest of the AI factories where we announced Novus, which is basically bringing just massive scale at 100 terabytes per second of aggregated throughput with a zettabyte scaled namespace. It is providing that fundamental underlying infrastructure for AI factories to be able to scale and ensure that the GPUs are doing what they do best. And they're not stalling. And all of the existing workloads don't go away for customers. They need to continue to be able to seamlessly deploy it. And the beauty of what we're enabling with the NetApp platform is for customers to have that unified data foundation that's bridging the gap across protocols, across on-prem and public cloud, and delivering that with one combined experience while making their data AI ready.

Rebecca Knight

>> So Helen, building on what Sandeep is saying about the importance of consistent data infrastructure, why, as companies are pushing AI workloads and now really we heard in the keynote every workload is AI, why has it become so much more important?

Helen Yu

>> Well, it's really important because consistency builds the trust, right? It's not just a technical trust. It is organizational trust. When teams really don't have the trust, they stop really making decisions, they start guessing, second guessing. And then that's where actually AI creates a value where people don't have to second guess. They just focus on getting things done. So to me, like consistent data structure is really the foundation, right? Combined with the accountability, you have to make sure you assign accountability that's aligned with the scope of delegation. And once that happens, that becomes the foundation for scalable and responsible AI deployment.

Christophe Bertrand

>> Right. And do you find that the organizations you engage with, either fractionally or full-time, are actually ready for these conversations? Or are they not?

Helen Yu

>> Well, some of them are. Some are still very siloed structure, right? We talk about fragmented organization. To me, it's not a storage problem. It's a leadership and accountability issue. When teams don't have access to consistent data, trusted data consistently, they actually stop making decisions. They start making workarounds. That's why shadow AI, shadow IT exists. Right. That's really the challenge we see in some of the organizations I work with. They focus more on technology, less on what business outcome we're trying to drive. What I would recommend is you start with the business outcome, right? What outcome you want to drive? And then you figure out what kind of operating model we need to have to make that work and then take that backward. And then you focus on how do I do engineering infrastructure from the back, backwards. That's how. You remove the silos.

Sandeep Singh

>> What's interesting, what Helen is highlighting is first, a focus on the right outcomes.

Helen Yu

>> Yeah.

Sandeep Singh

>> Define the problem statement and what the outcome you're trying to produce and then the technology is in service of that. So 100% agreed on that. And I think the other point you're reinforcing is you've got to really be able to think through the people, process, technology aspects. And then people, basically the organizational structure becomes another fundamental item that needs to be thought through here.

Helen Yu

>> Absolutely. Yeah. I think people, process, technology. I love that. Plus the planet, right? Yeah. Being more responsible.

Christophe Bertrand

>> Right. But that's only possible if you have a good understanding and management of the data estate.

Rebecca Knight

>> Right.

Christophe Bertrand

>> Which brings up a couple of other aspects of AI now being in the fold of helping you control and manage the environment.

Sandeep Singh

>> Yeah.

Christophe Bertrand

>> So I have a question for you, Sandeep, because there have been some announcements around you opening up a little more to maybe the ability for agents, for lack of a better term, to become part of that control plane.

Sandeep Singh

>> Yes.

Christophe Bertrand

>> Which of course for a lot of traditional storage people may be a big red flag, but in reality can be a great advantage provided the right guardrails are in place. Tell us more about that.

Sandeep Singh

>> Yeah, look, our vision is that customers obviously have distributed data estates. And not only do they need a consistent data plane, but they need a unified control plane. And the forward-looking future is not just humans managing infrastructure, it's basically agents along with humans there. So it has to be designed as an autonomous control plane. And as part of the announcements that we have made with NetApp Console, This is enabling this autonomous optimization future for customers. This is where we're designing and building agents to enable more intent-based outcomes. As Helen alluded to, outcomes is what matters. So enabling customers to be able to specify the outcomes that they want, enabling them to be able to specify across their fleet. Here's the policies and the policy boundaries. That they want to be able to set. And then with those intentions identified, then agents can be working continuously to maintain and optimize the infrastructure. A powerful example that was just showcased earlier in the keynote is this notion of agents that are set up. They, when, in any infrastructure, if and when bad things happen, they happen typically in the middle of the night people are woken up and if there are performance anomalies that happen instead of waking people up, agents can be monitoring as and when a performance anomaly starts to occur based on predefined boundaries and policies. Agents can in real time implement the necessary quality of service rules to ensure there's compliance to the boundary conditions there. And that way, no human is woken up. The right outcome is achieved and there's no noisy neighbor effect. And there's an audit trail presented to the humans for going and following up.

Rebecca Knight

>> It goes back to the importance of trust too. They actually believe that it is a truly autonomous operation.

Sandeep Singh

>> 100%. Absolutely. And this is where this notion of human in the loop is going to matter in terms of building that trust, coupled with the right policy boundaries and guardrails have been defined by the humans. And then there's trust that, yes, these boundaries and guardrails are going to be implemented and enforced. So, that really matters. But at the end of the day, human in the loop is going to be there for a long time.

Helen Yu

>> Speaking of trust, I think autonomous systems earn trust the same way people earn trust, right? What it does through the transparency, or demonstrated judgment and also boundary, right? Clear, clarified boundary. To me, those are very important. I would use a framework I call CDA framework. The first one is consistency. Can the organization really measure the behavior regardless of human or machine interacting with you? And then the D is decision quality, right? When you make a decision, Does that improve the outcome rather than just move things around? And then lastly, adaptability. When you're trying to get— can you adapt to changes quickly with all the automation or are you being locked in with the automation?

Sandeep Singh

>> Yeah, I think those are fantastic points. And one of the things you touched upon, machine, what we have to recognize is basically things are happening at machine speed, right? And you need these actions to be implementable or implemented at machine speed level in order to just keep everything continuously optimized. Absolutely. And look, decision logs, the audit trails are going to be an important part of, coupled with the necessary guardrails of just continuing to build that trust.

Helen Yu

>> Absolutely.

Rebecca Knight

>> But we've seen instances where those decision logs have been modified too, which we're not going to get into that. But I want to ask you, Helen, about measuring the success of automation because so often it's just, well, how many people did we get to remove? How many manual tasks are no longer in this process? What do you think leaders should be looking at when they're really trying to determine whether or not a process is genuinely improved?

Helen Yu

>> I would say you should put your technology aside first, right? Think about where you are at today, where you want to be, and then benchmark, right? we talk about start with business outcome. And what does that mean to finance versus marketing? It might be very different, or your technology team or customer team might be different, right? And then talk to your customer partners, right? Is that really aligned with what they're trying to achieve in collaboration with you? And then you can put in the measurable metrics, not just for humans, but also for agents. What does that mean, right, for agents to act autonomously? And then you want to make sure you have a governance model in place. To me, the, hey, algorithm decides is not a governance model. You need to really understand driving that accountability. We used to do the RACI chart, right, in services. Now you need to incorporate the machine or agents into your RACI chart. Who owns what data? Who has access to what data? And then when you're going to make an exception, who has the right to override the exception? And then make sure you understand what belongs to the machine, right? Repeatable work for machine to do, and then the policy-driven work for them to do, versus the human needs to be part of the judgment. So, if there's priority conflict or there's tacit contextual knowledge they have and then they need to get involved. So that's very important to me. It's like humans should, we should really focus shifting the repeatable work to machines, let humans own the consequential tasks.

Christophe Bertrand

>> Right. So you're making the case that autonomous operations heavily depend on context. There's no question about that. And it's not just a collection of tasks. It is truly autonomous. So Sandeep, from a control plane perspective now, leaving the data plane aside for a second, What do you need to know? What does the control plane need to understand to be able to contextualize the actual actions beyond the data component?

Sandeep Singh

>> Look, the control plane, first of all, needs to be able to provide that consistency across on-prem, across cloud, across a NeoCloud, because if you have fragmentation, then again, the context is lost. And it becomes siloed or fragmented. So that's kind of the first part of it. Secondly, the control plane needs to be able to understand what are the defined policies in a given organization or even within a given organization within a specific department or a function, because those policies can turn— can be different. So that needs to be provided, not on an incident-by-incident basis, but across that global fleet. And because there will be, that consistency that you do need. And that's kind of like the second piece. Thirdly, it becomes important as and when that control plane and agents are making decisions, what type of operations are they allowed to do and what are the operations they're not allowed to do. If there's any type of even metadata access, what can they access? What can they not access? So those need to be very clear as part of building trust.

Helen Yu

>> Yeah.

Sandeep Singh

>> And as part of those operations, then lastly becomes essentially what can they just act on versus doing all of the research and then presenting a recommendation to the human. So a perfect example is when you're provisioning the underlying data infrastructure. Today, a human will go through and there's a number of 30, 40 different decision points that someone might go through across their fleet to understand and deploy and provision the underlying infrastructure. Agents can autonomously go and take all of those repetitive tasks and be able to conduct them in parallel and come back with— across all of your entire fleet, here's a recommendation. Now the human in the loop can go and approve that part of it. So this is where those repetitive tasks can be removed, overall experience simplified, and now the humans are able to go and still maintain that trust boundary. And still benefit also from just that consistent operating experience.

Helen Yu

>> Yeah, I would agree. But the only thing I would add here is that it's not static. Things evolve, right? As things could change and policy could change

Sandeep Singh

>> and—100%.

Helen Yu

>> Yeah.

Rebecca Knight

>> No, absolutely. So as we're wrapping up here, we're all here at Insight 2026. What are some of the conversations that you're having today that you think will be the most resonant as you go forward in this new year, in this new AI era? Helen, I'm going to start with you.

Helen Yu

>> I am extremely energized walking away from the keynote yesterday and today. Honestly, I've been in tech for a long time, being with Oracle for 12 years, and worked there as a solution architect. this is the first time I feel like we have everything we need to build the most powerful system to really serve humankind from healthcare perspective, from financial services perspective, every single field you think about, right? And AI is going to play a major role. But it's not about the technology, it's about how we build it as humans, right? We should design the role, not around what automation can eliminate. We design the role around what humans can own. To me, that mindset is so critical. And then I was actually teaching a business class at Loyola Law Center last week. So the student asked me, hey, can you tell us what's the difference between automation and autonomy? So here's what I told them. Automation, think about a vending machine. You go there, you put in a coin, you press B7, a bag of chips comes out. And autonomy, if you hire a personal trainer, you tell the trainer, I want to get more fit. You define it, then you leave it to your trainer to decide how to, go through the training for you. And I think we're going through that stage right now. Everyone has different definitions about what being fit means. And then when you delegate the task, you are delegating expectation around outcome, right, or output. But when you delegate a decision, you're delegating the expectation around judgment. Right. So can we trust AI to make the judgment on our behalf? It's up to us. We're the ones who can make that happen and make it happen securely, responsibly in the longer term. That's my biggest takeaway from INSIGHT this week.

Rebecca Knight

>> How about you, Sandeep?

Sandeep Singh

>> Oh, look, it's clear that AI is going to be pervasive. It is pushing on all of the boundaries at the entire infrastructure level across the areas of performance and scalability, the governance aspects become incredibly important. Data ultimately is the fuel for AI, and data has become the most precious asset in this case. Fundamentally, Data not only needs to be performant, but it needs to provide governed access. And data is distributed as we began the conversation. It's the underlying consistent data infrastructure strategy becomes fundamentally important. Having a foundation that bridges across on-prem and cloud and Neo Clouds is the beginning building block. There's the— and combining that with the autonomous control plane that provides one operating experience. But more importantly, going beyond that is how do customers truly unlock the value of data? This is where bringing metadata for discovery as well as then being able to convert the data into AI-ready data and providing governed access without creating copies, while maintaining control is becoming so fundamentally important. And these are all of the areas that we're building with the NetApp platform for enabling customers to truly be able to lead in the AI era.

Rebecca Knight

>> NetApp is in a sweet spot. Sandeep and Helen, thank you both so much for coming on the show.

Christophe Bertrand

>> Thank you. Thank you.

Rebecca Knight

>> I'm Rebecca Knight with Christophe Bertrand. Stay tuned for more of theCUBE's live coverage of NetApp Insight. You're watching theCUBE. The leader in enterprise tech news and analysis.