Snowflake Summit 2026 | Mike Blandina, Snowflake & Vinayak Kagalkar, Clover

Clips
More from Snowflake Summit 2026

Mike Blandina

Chief Information Officer

Snowflake

Vinayak Kagalkar

SVP, CTIO/Head of Engineering

Clover

play_circle_outline Snowflake and Clover: Powering SMB Commerce with Integrated POS, Payments, Embedded Finance, and AI-Driven Analytics

play_circle_outline CIO role evolving: strategist to hands-on builder, building internal AI first

play_circle_outline Managing AI Adoption in Regulated Finance: Agents, CoCo, and Automated Controls for Compliance

play_circle_outline Standardize architecture while allowing model/cloud/tool flexibility for use cases

play_circle_outline Productivity gains from natural-language interfaces, CoCo desktop, and CLI resurgence

Info

Mike Blandina, Snowflake & Vinayak Kagalkar, Clover

Mike Blandina

Chief Information Officer Snowflake

Vinayak Kagalkar

SVP, CTIO/Head of Engineering Clover

search

Rebecca Knight

>> Good afternoon everyone, and welcome back to theCUBE's live coverage of the Snowflake Summit 2026 here at Moscone. I'm your host, Rebecca Knight alongside Dave Vellante, Co-Founder of theCUBE. We've got a terrific guest for this next segment. Mike Blandina, Chief Information Officer at Snowflake. Welcome back, Mike.

Mike Blandina

>> Yeah, thanks for having me.

Rebecca Knight

>> Great to have you.

Mike Blandina

>> Yeah.

Rebecca Knight

>> And Vinayak Kagalkar, SVP, CIO, Head of Engineering at Clover Fiserv. Welcome.

Vinayak Kagalkar

>> Thank you.

Rebecca Knight

>> I want to start with you, Vinayak, to tell our viewers a little bit about Clover, which is part of Fiserv. So tell us a little bit about what you do.

Vinayak Kagalkar

>> Yeah. As you notice, Clover is the financial and commercial business platform supporting the SMB-focused market. We are about 13 years, First Data acquired Clover around 2013, and been focused as a commerce platform for SMBs. We are serving about million merchants and 12 countries processing about 330 billion GPV over the years. And the special about Clover is Clover is fully integrated commerce platform. We build our purpose-built hardware fully integrated with payment processing as well as point of sale and also we provide the back office embedded finance and that's the core part of it's a well integrated system. Historically, these integrations was available only for enterprise merchants. The focus here is to make all these fully integrated features to our SMBs, and that's one of the areas where we partnered with Snowflake about eight years back. We are one of the earlier partners with Snowflake, and the purpose there was how can we make the intelligent data access visible and available to SMB merchants? That was the focus where we started, and we chose Snowflake as a partner, not only as a data warehouse, it's a data platform. And that time, the AI was not there, but we wanted to make sure that we provide business analytics and insights to the SMB merchants, what is typically available for enterprise. From there, we started the journey, and today we are with AI era, ML. We started using building agents so that way we can build business insight agents so that SMB merchants can understand their business very well, and also we can provide actionable insights to them so that it can make the necessary changes to improve their business and that's where we're.

Rebecca Knight

>> Great.

Dave Vellante

>> Okay. So this is the CIO segment, right? That's why you guys are paired up here.

Mike Blandina

>> Yes.

Dave Vellante

>> Okay. What has the CIO got on his or her plate today?

Mike Blandina

>> Well, depending on the day, I can go anywhere from strategist to janitor and then sometimes sales in between. So in our show here, I'm a little bit of everything and above, but certainly at Snowflake, we're trying to stay ahead of our customers, but also stay connected with them, so building great AI capability on top of our data cloud and building things. My team is building things internally that we know will be relevant for our customers. So I'm building them for Snowflake first, and then some of those things make them their way into the product so that our customers don't have to build them. They become some of the standard features when the product is delivered.

Dave Vellante

>> It's interesting. I was interviewed, Bipul Sinha, who's the CEO of Rubrik at RSAC and he said something similar, Mike. He said, "You want to be ahead of your customers, but you don't want to be 200 miles ahead of your customers."

Mike Blandina

>> Right.

Dave Vellante

>> "You want to be close enough that you can solve the problems before they do, but in close proximity to that problem." Do you find presumably the same thing?

Vinayak Kagalkar

>> Yeah, same thing. We started using AI and data analytics for our own internal business operation, right? How we are doing in terms of our feature adaptation, our customer metrics, our business metrics, revenue and supply chain. And that's where we started building our own internal toolings, and then we started opening those features to our merchants since we are in the merchant solution so we provide.

Rebecca Knight

>> So as CIOs, how does how you spend your time today differ from how you spent your time say five years ago before this AI boom?

Mike Blandina

>> I'll start. Five years ago, I wasn't vibe coding on the weekend.

Rebecca Knight

>> Okay. All right, there you go.

Mike Blandina

>> And so that's a new thing.

Dave Vellante

>> Not even surprised.

Mike Blandina

>> But I do find myself beyond the vibe coding piece of it, really making sure we understand the impact our tools are having, and then how can I deploy those internally, both help the company inside but help the product going outside, and it's a different mindset really. So as an example, five years ago, I would ask my team to write a product spec. Today, I asked them to type their thoughts into CoCo because CoCo will turn that into a markdown document, and from that markdown document, you can immediately start building capabilities. And that's just a completely different way to think about the problem, so that's a simple example.

Rebecca Knight

>> Yeah. How about you, Vinayak?

Vinayak Kagalkar

>> And the same thing as Mike said, right? We started vibe coding. I used to look for front end developers to carve out the thoughts I have in mind. Now I don't need that. I can just go and create vibe code and prompt it in vibe code and I can tell my story. This is what I want to build. That's becoming easy now. Same time telling the story, like use cases, telling the story to the product team or sales team, that's become easy now. That's one part of it. The second part is the operational aspects of it. Historically, I saw somewhere we saw no Jira tickets anymore. We don't need to ask someone to do the work. We are very self-service now. We can build a self-service agent. We are able to do our work with the agent, or we can quickly access the data now with these agents and tools. I see the difference in day-to-day power.

Dave Vellante

>> Interesting. So I'm interested in what you're vibe coding. I'll vibe code stuff that, A, doesn't matter or, B, is like a prototype. What are you guys vibe coding?

Mike Blandina

>> Well, part of what I do is I'm trying to experience what my team's experiencing. And so I gave a team, we were looking at thinking about building a new pricing and billing platform, and I was getting a bunch of pushback from my team. "Well, you got to integrate it with this and you got to change this." And they were looking from the paradigm of what they built five years ago, and I said, "It's okay to burn that down. Let's think about it brand new because we can do." In years past with the old development life cycle, you might remodel versus build new-

Dave Vellante

>> Right....

Mike Blandina

>> because the remodel was cheaper or quicker. And so I challenged them and said, "I'm going to go try it myself. I've built this before many years ago. Let me just start going at it." And so I would just carve out four hours every Sunday because that's what my wife would give me. And in five Sundays, I built not just a prototype, I built a full working system, 60 entities, a full React front end, all the APIs, a full batch process. And then I said, "You do a demo of yours and I'll do a demo of mine." And by the way, they had some terrific things they built as well, right? But it was a way for me to better understand what they were experiencing and what's working, what's not working. And we also learned from it to build new features into the tools.

Rebecca Knight

>> And yet it did need to come from the top of it's okay to put that in the-

Mike Blandina

>> No question.

Rebecca Knight

>> To do away with that old system of doing it.

Mike Blandina

>> So I literally sat in a room and said, "Burn it down. You have my permission." So yeah.

Rebecca Knight

>> Excellent.

Dave Vellante

>> So change management is such an important part of the CIO's role and feels like a lot of organizations are going through a lot of change. How would you describe the dynamics today in your organization, and how you're handling the change management?

Vinayak Kagalkar

>> Yeah, no, that's an interesting part. And especially we are in a financial services industry, the change is very difficult because of the compliance and regulatory aspects of it. But I think that industry has accepted now and especially in the engineering force, people started accepting the benefits of the AI and automation part of it. There's a lot of places where historically, we used to look for code reviews, SDLC processes used to take two to three weeks to make a simple changes. Now, we are able to do it within a half day and commit it, right? I think people have started seeing the advantages of it and people started accepting it. But same time, I think there is still a control freakness is still there, right? The controls like, are we ready to put this code in front of customers? I think since financial industry has a good control system, I think we apply the same controls for AI agents and tools. We feel good about it. I think I see significant adaptation last six months in AI tooling and all this.

Dave Vellante

>> So as a software company, you can presumably move faster than a highly regulated industry, but does the pace of change negatively affect your ability to introduce or enforce standards, and how you deal with that?

Mike Blandina

>> Well, I think you can enforce standards in the model. As an example in CoCo, you can just say in this particular project, anything we build, here's a set of boundaries that you as my agent for code development should not cross. Here's my standards for the type of artifacts we want to use and the type of code we want to use, and so you can do that upfront just like you would do on a paper document and enforce through the software development life cycle. We've done things like in my team, we moved to a one-week Agile sprint cycle and just that idea really, people's heads exploded the first month or two that we did that, and now everyone looks back four or five months later and goes, "Actually, it works and it makes complete sense." It's just moving faster, but we can still do our ceremonies. We can still do our processes, but the actual coding part of it just goes so much faster.

Dave Vellante

>> So what was the sprint prior to that? What was the duration of sprint?

Mike Blandina

>> Three weeks.

Dave Vellante

>> Three, even that's pretty fast, Mike.

Rebecca Knight

>> Be ambitious.

Mike Blandina

>> Yes. Yeah. Most of the teams I've worked with in the past when I was at JP Morgan, we were on a two-week sprint cycle. And so we had agreed when I joined Snowflake to do three, but we went from three to one and it was a shock. It was a shock to the system, but now it's like... And again, your ceremonies, your demos go much quicker, your planning discussions go much quicker, and you need less time for the coding part of the sprint, right? So it makes a ton of sense, but these are all scary changes for people that have been doing something a certain way for 20 years.

Dave Vellante

>> Well, you mentioned CoCo as a way to enforce those standards. Here's your boundaries go. That assumes everybody uses CoCo. Now CoCo, I haven't used it, but it sounds pretty good so people like it and do it. You mentioned JPMC. You had Jamie Dimon say, "Thou shalt embrace AI."

Mike Blandina

>> Yes.

Dave Vellante

>> "Or you're in trouble." Okay, great. And people at the organization probably said, "Okay, let's go." And then now you have a lot of things-

Mike Blandina

>> When Jamie speaks, people listen.

Rebecca Knight

>> Yes. He has that effect.

Dave Vellante

>> Even people outside of JP Morgan Chase.

Mike Blandina

>> Yes.

Dave Vellante

>> But you have a lot of activity bubbling up now. So you had the top down edict and now you got the personal innovation bubbling up. How are you seeing that in your organization, number one? Number two, how do you then make sure you're not just creating islands of innovation?

Vinayak Kagalkar

>> Yeah. I think we started initially-

Dave Vellante

>> Or I should say islands of intelligence is a better term. Right, go ahead.

Vinayak Kagalkar

>> Yeah, I think initially, it was a curiosity, right? Everybody was trying to vibe code, everybody was trying to use the AI tools, but now I think as the industry has matured, the tooling has come to surface. People started accepting it. I think the main thing is from architecturally, we still have the old principles, right? I think as part of the evolution of data platform building, we try to bring the data into centralized warehouse, data in the silos has been cleaned up. I think those principles are still there, and typically this is helping a lot for AI story. Historically, we had a databases with their own components and other stuff. We built everything in a Snowflake base, and then we started putting agents to help this. That's the one couple of things to help. But I think there are, people are still trying different models that's still there. There are people are still trying different harnesses, whether it's a cloud, OpenAI, Gemini. And there are some engineers are attached emotionally with one tool versus another tool, but we are trying to standardize that as go, but at least we want to give flexibility for our engineering teams to build the right things. But same time, architecturally, we are standardizing how we use data, whereas an architectural principles we do. And as Mike said, try to put some controls around it. We don't want to go and get some middleware randomly, 25 middlewares floating around. We try to standardize like all user interfaces should be in React if it is a microservice framework, should be in Java or Springboard because it's the long run, we still need to maintain the code and manage it. So that guidelines we put it, but same time this industry is evolving, the tools are evolving. We need to be open-minded to adopt. And then based on the use cases, right, because sometime if you're handling a particular data, maybe particular models are good. If you're handling a video or a multimedia contents, maybe the other models may be good. So we're a little bit open on that aspect, but same time, you don't want to compromise on the architectural principles.

Dave Vellante

>> But you had the latitude to do that at Clover and you could pick and choose the pieces of Fiserv that you wanted to bring in. And in many respects, Snowflake, you had the advantage because you got the Snowflake data platform. For a larger company, they've got a lot of everything.

Mike Blandina

>> Right.

Dave Vellante

>> And the different departments have their own data sets. How would you, thinking about maybe your experience at JPMC, a lot of this is actually, they'll talk about publicly, so I'm not asking for confidential information, but just generically, a large organization like that that's very complex, how do they move from, which is a CIO challenge, point A to where they want to get without having to bring in an army of consultants, maybe you need some consultants, but how do you approach that, Mike? Do you start at the edge? Do you have Skunk Works and then learn from that? What's the best practice that you would recommend there for large organizations?

Mike Blandina

>> Yeah. I think one of the things we see is solve what's very solvable today. Don't overreach. Take your most repetitive task solve them and AI can do that really well, really quickly. Inevitably, I think you've gotten to this. There is some culture in this, right, and there is a culture change, and we've seen some people take the AI red pill really quickly and others not swallow it very well. And I don't even think it's an age thing or... I do think some of your more senior engineers are steeped in the dogma of software development life cycles and some of them may feel less valuable just writing in English. I think early on, it's sort of a two sticks and a carrot approach, and eventually you get to one stick and one, a stick and a carrot. We've had to push it in terms of we're going to measure this. It becomes part of what you do, but eventually, I think most people get over the hump. Not all, but most get over the hump, and then you stop measuring and you start encouraging and you start rewarding based on behavior. And so you can start to see the teams that are really, really going fast and the ones that are still struggling a little bit because they're doing it the old way.

Dave Vellante

>> Am I right? You start by measuring and monitoring behavior and then you're measuring outcomes. Is that right?

Mike Blandina

>> Absolutely, yep. Behavior's got to come first because I think everyone's exploring and learning, right? And so we had the privilege of taking most of the company and saying, "Go do this," including our sales team. We said, "Go code in CoCo." And man, they built thousands of streamlined dashboards because that was the obvious thing they understood.

Dave Vellante

>> Yeah, look what I did.

Mike Blandina

>> Yeah, look what I did.

Dave Vellante

>> Totally get it.

Mike Blandina

>> And by the way, they all thought they were amazing. But then we then came back and reigned in the chaos a little bit, but it was good to give everyone a taste of it. And it taught our finance team what it meant to build with AI. And now, when their engineering partners are meeting with them, they're much better at understanding the process and why certain questions are getting asked and how to partner with them.

Rebecca Knight

>> You're describing something so interesting because AI just as a topic is pretty divisive right now in our country, in the world. And you're talking about changing behaviors. You're obviously a technologist and a tech optimist. How do you encourage people to think about AI? I mean, there's a lot of fear mongering going on, but is it just a tool? Is it just another tool that you now have to work with, or is it so revolutionary, which is what scares people, I think, and maybe why there's resistance.

Mike Blandina

>> I think today, it is both revolutionary and a tool.

Rebecca Knight

>> Okay, right.

Mike Blandina

>> But if you go... Things in technology repeat themselves, right? I'm old enough to remember the case tool days in the '80s, the 4GL days, and then more recently in the late 2000s, we had the low code, no code craze, right? I think it's certainly from a coding perspectives, this is the end evolution of all of that, right? And so I don't think you can ignore it, but we've all been there. I had this discussion with my wife last week where she said, I asked her a question and she gave me an answer and I said, "Well, that's not correct." And she said, "Well, that's what Gemini said." And I said, "I'm pretty sure Gemini's wrong." Right?

Rebecca Knight

>> Yeah.

Mike Blandina

>> But she is already, she's not an AI person or a tech person, but she now believes that if Gemini told her, it must be written in stone, right? And so I just think we're all, both families, work, coworkers, the entire industries are going through this same wave together.

Rebecca Knight

>> Yeah. It's a lot of change, a lot of change. And so how about you, Vinayak? When you are thinking about what you are experiencing here at Summit, what you're hearing in terms of new announcements, new innovations, the conversations that you're having with your peers and maybe your competitors too, what are you taking away that you're going to bring back to work with you on Monday? What is most striking to you?

Vinayak Kagalkar

>> I think as Mike said, there is a evolution as well as this is the way we need to adapt to it. And I think initially, there will be resistance, there will be a question whether this is a technology, this is going to work or not. But I think people have seen, at least now last couple of years, people have seen AI is working as we said, right? Again, you have various examples, people creating contents, people analyzing the data, whether it's a healthcare or the industry is autonomous driving, you name it. There's the various spaces we have seen the AI's presence. I think in especially software engineering space, it's going to be quickly adopted compared to any other industry, but I think there is a lot of productivity improvements that are happening. Of course, the token discussion is triggering now last couple of weeks, but having said that, if you can wisely use this technology, I think there's a lot of productivity gain is going to happen. We're going to produce the best in class products, but at the end of the day, we are going to produce product for the consumers who can use it and make their missions right. That's how we are in the business. And I see that. And I said, as Mike said also, we started with the VI editors, now the CLI is back, everybody's tacking command lines, right? Everybody wants to talk using command line. In between, we all gone through.

Mike Blandina

>> We did just announce the CoCo desktop.

Rebecca Knight

>> There you go.

Mike Blandina

>> You don't have to use the command line anymore.

Vinayak Kagalkar

>> But I think it's amazing how things are getting. It's using natural language, you can interact with the system and get the things done. I think people are accepting even always the various teams within that people are doing skills, people are writing core routines, right? Since I'm also involved in hardware, people are doing firmware updates, diagnostic using AI, people are writing those type of functionalities. It's a repeatable task, but how can they quickly do it? I think it's going to make a significant impact in the productivity.

Rebecca Knight

>> I think you're right. Vinayak, Mike, thank you both so much for coming on the show.

Dave Vellante

>> Thanks, guys.

Vinayak Kagalkar

>> Thank you, sure.

Mike Blandina

>> It was a pleasure, enjoyed it.

Vinayak Kagalkar

>> Bye now.

Rebecca Knight

>> I'm Rebecca Knight for Dave Vellante, stay tuned for more of theCUBE's live coverage of the Snowflake Summit. You're watching theCUBE, the leader in enterprise tech news and analysis.

Mike Blandina, Snowflake & Vinayak Kagalkar, Clover

search

Rebecca Knight

>> Good afternoon everyone, and welcome back to theCUBE's live coverage of the Snowflake Summit 2026 here at Moscone. I'm your host, Rebecca Knight alongside Dave Vellante, Co-Founder of theCUBE. We've got a terrific guest for this next segment. Mike Blandina, Chief Information Officer at Snowflake. Welcome back, Mike.

Mike Blandina

>> Yeah, thanks for having me.

Rebecca Knight

>> Great to have you.

Mike Blandina

>> Yeah.

Rebecca Knight

>> And Vinayak Kagalkar, SVP, CIO, Head of Engineering at Clover Fiserv. Welcome.

Vinayak Kagalkar

>> Thank you.

Rebecca Knight

>> I want to start with you, Vinayak, to tell our viewers a little bit about Clover, which is part of Fiserv. So tell us a little bit about what you do.

Vinayak Kagalkar

>> Yeah. As you notice, Clover is the financial and commercial business platform supporting the SMB-focused market. We are about 13 years, First Data acquired Clover around 2013, and been focused as a commerce platform for SMBs. We are serving about million merchants and 12 countries processing about 330 billion GPV over the years. And the special about Clover is Clover is fully integrated commerce platform. We build our purpose-built hardware fully integrated with payment processing as well as point of sale and also we provide the back office embedded finance and that's the core part of it's a well integrated system. Historically, these integrations was available only for enterprise merchants. The focus here is to make all these fully integrated features to our SMBs, and that's one of the areas where we partnered with Snowflake about eight years back. We are one of the earlier partners with Snowflake, and the purpose there was how can we make the intelligent data access visible and available to SMB merchants? That was the focus where we started, and we chose Snowflake as a partner, not only as a data warehouse, it's a data platform. And that time, the AI was not there, but we wanted to make sure that we provide business analytics and insights to the SMB merchants, what is typically available for enterprise. From there, we started the journey, and today we are with AI era, ML. We started using building agents so that way we can build business insight agents so that SMB merchants can understand their business very well, and also we can provide actionable insights to them so that it can make the necessary changes to improve their business and that's where we're.

Rebecca Knight

>> Great.

Dave Vellante

>> Okay. So this is the CIO segment, right? That's why you guys are paired up here.

Mike Blandina

>> Yes.

Dave Vellante

>> Okay. What has the CIO got on his or her plate today?

Mike Blandina

>> Well, depending on the day, I can go anywhere from strategist to janitor and then sometimes sales in between. So in our show here, I'm a little bit of everything and above, but certainly at Snowflake, we're trying to stay ahead of our customers, but also stay connected with them, so building great AI capability on top of our data cloud and building things. My team is building things internally that we know will be relevant for our customers. So I'm building them for Snowflake first, and then some of those things make them their way into the product so that our customers don't have to build them. They become some of the standard features when the product is delivered.

Dave Vellante

>> It's interesting. I was interviewed, Bipul Sinha, who's the CEO of Rubrik at RSAC and he said something similar, Mike. He said, "You want to be ahead of your customers, but you don't want to be 200 miles ahead of your customers."

Mike Blandina

>> Right.

Dave Vellante

>> "You want to be close enough that you can solve the problems before they do, but in close proximity to that problem." Do you find presumably the same thing?

Vinayak Kagalkar

>> Yeah, same thing. We started using AI and data analytics for our own internal business operation, right? How we are doing in terms of our feature adaptation, our customer metrics, our business metrics, revenue and supply chain. And that's where we started building our own internal toolings, and then we started opening those features to our merchants since we are in the merchant solution so we provide.

Rebecca Knight

>> So as CIOs, how does how you spend your time today differ from how you spent your time say five years ago before this AI boom?

Mike Blandina

>> I'll start. Five years ago, I wasn't vibe coding on the weekend.

Rebecca Knight

>> Okay. All right, there you go.

Mike Blandina

>> And so that's a new thing.

Dave Vellante

>> Not even surprised.

Mike Blandina

>> But I do find myself beyond the vibe coding piece of it, really making sure we understand the impact our tools are having, and then how can I deploy those internally, both help the company inside but help the product going outside, and it's a different mindset really. So as an example, five years ago, I would ask my team to write a product spec. Today, I asked them to type their thoughts into CoCo because CoCo will turn that into a markdown document, and from that markdown document, you can immediately start building capabilities. And that's just a completely different way to think about the problem, so that's a simple example.

Rebecca Knight

>> Yeah. How about you, Vinayak?

Vinayak Kagalkar

>> And the same thing as Mike said, right? We started vibe coding. I used to look for front end developers to carve out the thoughts I have in mind. Now I don't need that. I can just go and create vibe code and prompt it in vibe code and I can tell my story. This is what I want to build. That's becoming easy now. Same time telling the story, like use cases, telling the story to the product team or sales team, that's become easy now. That's one part of it. The second part is the operational aspects of it. Historically, I saw somewhere we saw no Jira tickets anymore. We don't need to ask someone to do the work. We are very self-service now. We can build a self-service agent. We are able to do our work with the agent, or we can quickly access the data now with these agents and tools. I see the difference in day-to-day power.

Dave Vellante

>> Interesting. So I'm interested in what you're vibe coding. I'll vibe code stuff that, A, doesn't matter or, B, is like a prototype. What are you guys vibe coding?

Mike Blandina

>> Well, part of what I do is I'm trying to experience what my team's experiencing. And so I gave a team, we were looking at thinking about building a new pricing and billing platform, and I was getting a bunch of pushback from my team. "Well, you got to integrate it with this and you got to change this." And they were looking from the paradigm of what they built five years ago, and I said, "It's okay to burn that down. Let's think about it brand new because we can do." In years past with the old development life cycle, you might remodel versus build new-

Dave Vellante

>> Right....

Mike Blandina

>> because the remodel was cheaper or quicker. And so I challenged them and said, "I'm going to go try it myself. I've built this before many years ago. Let me just start going at it." And so I would just carve out four hours every Sunday because that's what my wife would give me. And in five Sundays, I built not just a prototype, I built a full working system, 60 entities, a full React front end, all the APIs, a full batch process. And then I said, "You do a demo of yours and I'll do a demo of mine." And by the way, they had some terrific things they built as well, right? But it was a way for me to better understand what they were experiencing and what's working, what's not working. And we also learned from it to build new features into the tools.

Rebecca Knight

>> And yet it did need to come from the top of it's okay to put that in the-

Mike Blandina

>> No question.

Rebecca Knight

>> To do away with that old system of doing it.

Mike Blandina

>> So I literally sat in a room and said, "Burn it down. You have my permission." So yeah.

Rebecca Knight

>> Excellent.

Dave Vellante

>> So change management is such an important part of the CIO's role and feels like a lot of organizations are going through a lot of change. How would you describe the dynamics today in your organization, and how you're handling the change management?

Vinayak Kagalkar

>> Yeah, no, that's an interesting part. And especially we are in a financial services industry, the change is very difficult because of the compliance and regulatory aspects of it. But I think that industry has accepted now and especially in the engineering force, people started accepting the benefits of the AI and automation part of it. There's a lot of places where historically, we used to look for code reviews, SDLC processes used to take two to three weeks to make a simple changes. Now, we are able to do it within a half day and commit it, right? I think people have started seeing the advantages of it and people started accepting it. But same time, I think there is still a control freakness is still there, right? The controls like, are we ready to put this code in front of customers? I think since financial industry has a good control system, I think we apply the same controls for AI agents and tools. We feel good about it. I think I see significant adaptation last six months in AI tooling and all this.

Dave Vellante

>> So as a software company, you can presumably move faster than a highly regulated industry, but does the pace of change negatively affect your ability to introduce or enforce standards, and how you deal with that?

Mike Blandina

>> Well, I think you can enforce standards in the model. As an example in CoCo, you can just say in this particular project, anything we build, here's a set of boundaries that you as my agent for code development should not cross. Here's my standards for the type of artifacts we want to use and the type of code we want to use, and so you can do that upfront just like you would do on a paper document and enforce through the software development life cycle. We've done things like in my team, we moved to a one-week Agile sprint cycle and just that idea really, people's heads exploded the first month or two that we did that, and now everyone looks back four or five months later and goes, "Actually, it works and it makes complete sense." It's just moving faster, but we can still do our ceremonies. We can still do our processes, but the actual coding part of it just goes so much faster.

Dave Vellante

>> So what was the sprint prior to that? What was the duration of sprint?

Mike Blandina

>> Three weeks.

Dave Vellante

>> Three, even that's pretty fast, Mike.

Rebecca Knight

>> Be ambitious.

Mike Blandina

>> Yes. Yeah. Most of the teams I've worked with in the past when I was at JP Morgan, we were on a two-week sprint cycle. And so we had agreed when I joined Snowflake to do three, but we went from three to one and it was a shock. It was a shock to the system, but now it's like... And again, your ceremonies, your demos go much quicker, your planning discussions go much quicker, and you need less time for the coding part of the sprint, right? So it makes a ton of sense, but these are all scary changes for people that have been doing something a certain way for 20 years.

Dave Vellante

>> Well, you mentioned CoCo as a way to enforce those standards. Here's your boundaries go. That assumes everybody uses CoCo. Now CoCo, I haven't used it, but it sounds pretty good so people like it and do it. You mentioned JPMC. You had Jamie Dimon say, "Thou shalt embrace AI."

Mike Blandina

>> Yes.

Dave Vellante

>> "Or you're in trouble." Okay, great. And people at the organization probably said, "Okay, let's go." And then now you have a lot of things-

Mike Blandina

>> When Jamie speaks, people listen.

Rebecca Knight

>> Yes. He has that effect.

Dave Vellante

>> Even people outside of JP Morgan Chase.

Mike Blandina

>> Yes.

Dave Vellante

>> But you have a lot of activity bubbling up now. So you had the top down edict and now you got the personal innovation bubbling up. How are you seeing that in your organization, number one? Number two, how do you then make sure you're not just creating islands of innovation?

Vinayak Kagalkar

>> Yeah. I think we started initially-

Dave Vellante

>> Or I should say islands of intelligence is a better term. Right, go ahead.

Vinayak Kagalkar

>> Yeah, I think initially, it was a curiosity, right? Everybody was trying to vibe code, everybody was trying to use the AI tools, but now I think as the industry has matured, the tooling has come to surface. People started accepting it. I think the main thing is from architecturally, we still have the old principles, right? I think as part of the evolution of data platform building, we try to bring the data into centralized warehouse, data in the silos has been cleaned up. I think those principles are still there, and typically this is helping a lot for AI story. Historically, we had a databases with their own components and other stuff. We built everything in a Snowflake base, and then we started putting agents to help this. That's the one couple of things to help. But I think there are, people are still trying different models that's still there. There are people are still trying different harnesses, whether it's a cloud, OpenAI, Gemini. And there are some engineers are attached emotionally with one tool versus another tool, but we are trying to standardize that as go, but at least we want to give flexibility for our engineering teams to build the right things. But same time, architecturally, we are standardizing how we use data, whereas an architectural principles we do. And as Mike said, try to put some controls around it. We don't want to go and get some middleware randomly, 25 middlewares floating around. We try to standardize like all user interfaces should be in React if it is a microservice framework, should be in Java or Springboard because it's the long run, we still need to maintain the code and manage it. So that guidelines we put it, but same time this industry is evolving, the tools are evolving. We need to be open-minded to adopt. And then based on the use cases, right, because sometime if you're handling a particular data, maybe particular models are good. If you're handling a video or a multimedia contents, maybe the other models may be good. So we're a little bit open on that aspect, but same time, you don't want to compromise on the architectural principles.

Dave Vellante

>> But you had the latitude to do that at Clover and you could pick and choose the pieces of Fiserv that you wanted to bring in. And in many respects, Snowflake, you had the advantage because you got the Snowflake data platform. For a larger company, they've got a lot of everything.

Mike Blandina

>> Right.

Dave Vellante

>> And the different departments have their own data sets. How would you, thinking about maybe your experience at JPMC, a lot of this is actually, they'll talk about publicly, so I'm not asking for confidential information, but just generically, a large organization like that that's very complex, how do they move from, which is a CIO challenge, point A to where they want to get without having to bring in an army of consultants, maybe you need some consultants, but how do you approach that, Mike? Do you start at the edge? Do you have Skunk Works and then learn from that? What's the best practice that you would recommend there for large organizations?

Mike Blandina

>> Yeah. I think one of the things we see is solve what's very solvable today. Don't overreach. Take your most repetitive task solve them and AI can do that really well, really quickly. Inevitably, I think you've gotten to this. There is some culture in this, right, and there is a culture change, and we've seen some people take the AI red pill really quickly and others not swallow it very well. And I don't even think it's an age thing or... I do think some of your more senior engineers are steeped in the dogma of software development life cycles and some of them may feel less valuable just writing in English. I think early on, it's sort of a two sticks and a carrot approach, and eventually you get to one stick and one, a stick and a carrot. We've had to push it in terms of we're going to measure this. It becomes part of what you do, but eventually, I think most people get over the hump. Not all, but most get over the hump, and then you stop measuring and you start encouraging and you start rewarding based on behavior. And so you can start to see the teams that are really, really going fast and the ones that are still struggling a little bit because they're doing it the old way.

Dave Vellante

>> Am I right? You start by measuring and monitoring behavior and then you're measuring outcomes. Is that right?

Mike Blandina

>> Absolutely, yep. Behavior's got to come first because I think everyone's exploring and learning, right? And so we had the privilege of taking most of the company and saying, "Go do this," including our sales team. We said, "Go code in CoCo." And man, they built thousands of streamlined dashboards because that was the obvious thing they understood.

Dave Vellante

>> Yeah, look what I did.

Mike Blandina

>> Yeah, look what I did.

Dave Vellante

>> Totally get it.

Mike Blandina

>> And by the way, they all thought they were amazing. But then we then came back and reigned in the chaos a little bit, but it was good to give everyone a taste of it. And it taught our finance team what it meant to build with AI. And now, when their engineering partners are meeting with them, they're much better at understanding the process and why certain questions are getting asked and how to partner with them.

Rebecca Knight

>> You're describing something so interesting because AI just as a topic is pretty divisive right now in our country, in the world. And you're talking about changing behaviors. You're obviously a technologist and a tech optimist. How do you encourage people to think about AI? I mean, there's a lot of fear mongering going on, but is it just a tool? Is it just another tool that you now have to work with, or is it so revolutionary, which is what scares people, I think, and maybe why there's resistance.

Mike Blandina

>> I think today, it is both revolutionary and a tool.

Rebecca Knight

>> Okay, right.

Mike Blandina

>> But if you go... Things in technology repeat themselves, right? I'm old enough to remember the case tool days in the '80s, the 4GL days, and then more recently in the late 2000s, we had the low code, no code craze, right? I think it's certainly from a coding perspectives, this is the end evolution of all of that, right? And so I don't think you can ignore it, but we've all been there. I had this discussion with my wife last week where she said, I asked her a question and she gave me an answer and I said, "Well, that's not correct." And she said, "Well, that's what Gemini said." And I said, "I'm pretty sure Gemini's wrong." Right?

Rebecca Knight

>> Yeah.

Mike Blandina

>> But she is already, she's not an AI person or a tech person, but she now believes that if Gemini told her, it must be written in stone, right? And so I just think we're all, both families, work, coworkers, the entire industries are going through this same wave together.

Rebecca Knight

>> Yeah. It's a lot of change, a lot of change. And so how about you, Vinayak? When you are thinking about what you are experiencing here at Summit, what you're hearing in terms of new announcements, new innovations, the conversations that you're having with your peers and maybe your competitors too, what are you taking away that you're going to bring back to work with you on Monday? What is most striking to you?

Vinayak Kagalkar

>> I think as Mike said, there is a evolution as well as this is the way we need to adapt to it. And I think initially, there will be resistance, there will be a question whether this is a technology, this is going to work or not. But I think people have seen, at least now last couple of years, people have seen AI is working as we said, right? Again, you have various examples, people creating contents, people analyzing the data, whether it's a healthcare or the industry is autonomous driving, you name it. There's the various spaces we have seen the AI's presence. I think in especially software engineering space, it's going to be quickly adopted compared to any other industry, but I think there is a lot of productivity improvements that are happening. Of course, the token discussion is triggering now last couple of weeks, but having said that, if you can wisely use this technology, I think there's a lot of productivity gain is going to happen. We're going to produce the best in class products, but at the end of the day, we are going to produce product for the consumers who can use it and make their missions right. That's how we are in the business. And I see that. And I said, as Mike said also, we started with the VI editors, now the CLI is back, everybody's tacking command lines, right? Everybody wants to talk using command line. In between, we all gone through.

Mike Blandina

>> We did just announce the CoCo desktop.

Rebecca Knight

>> There you go.

Mike Blandina

>> You don't have to use the command line anymore.

Vinayak Kagalkar

>> But I think it's amazing how things are getting. It's using natural language, you can interact with the system and get the things done. I think people are accepting even always the various teams within that people are doing skills, people are writing core routines, right? Since I'm also involved in hardware, people are doing firmware updates, diagnostic using AI, people are writing those type of functionalities. It's a repeatable task, but how can they quickly do it? I think it's going to make a significant impact in the productivity.

Rebecca Knight

>> I think you're right. Vinayak, Mike, thank you both so much for coming on the show.

Dave Vellante

>> Thanks, guys.

Vinayak Kagalkar

>> Thank you, sure.

Mike Blandina

>> It was a pleasure, enjoyed it.

Vinayak Kagalkar

>> Bye now.

Rebecca Knight

>> I'm Rebecca Knight for Dave Vellante, stay tuned for more of theCUBE's live coverage of the Snowflake Summit. You're watching theCUBE, the leader in enterprise tech news and analysis.