Making AI Real for the Enterprise: Inside Snowflake’s AI Strategy

Speaker A: Welcome to the Uncharted podcast. I am so excited to have this guest with me, Bala Kasi Gvasanathan. It's so good to have you on the Uncharted podcast.

Speaker B: Thank you so much for having me. And more importantly, thank you for doing all that you do for founders and entrepreneurs through your podcast.

Speaker A: Let's start with a very quick intro bio. I'd love the listeners to understand a little more context on who Bala is.

Speaker B: I set up AI and developer experiences here at Snowflake. And what that means is really in terms of helping developers and builders turn AI into something useful and drive outcomes. And the two parts I care deeply about is what are we helping users in Snowflake? Uh, build in Snowflake. That includes things like notebooks, streamlit apps, and things of that nature, as well as build on Snowflake because we really want developers and partners to build stuff. So it's all about how do we build the best platform for everybody there. Prior to this, I've been spending a bunch of time at Microsoft, then with Apigee, which got acquired by Google Cloud, and then of course spent an, uh, interesting couple of years at Google X as well, and now at Snowflake. One thing you'll appreciate is I actually, interesting fact, have been doing go to market roles, pretty much everything you can imagine. Hands m on as a leader or as a leader, be it bdr, salesperson, product marketing, sales ops, customer experience. Each one of those. And that, I think shapes you as a product leader.

Speaker A: Well, you've gone through a couple of, let's call it, technology transformations. You were at Microsoft when things went from PC to mobile, apigee with app integrations and whatnot. And now it's Snowflake, but let's call it the AI transformation. What are some of the commonalities between those and what makes this one as it pertains to the AI, uh, transformation different than the first two?

Speaker B: Yeah, absolutely grateful. Like you said, been through client, server, cloud, mobile, and now AI and data. Right. I couldn't have wished for a better learning and experience path than that. I think the way I think about this is two things, right? What hasn't changed is I think you're still. You have to remind yourself that you're here to solve the problems for customers and you're to solve it in the right way and for the right things. That doesn't change. I think the way in which you do it and how you do it changes, but the core does not change. And you need to drive value for the customers. What has definitely Changed is the ability to deliver things faster and learn faster. So I think we are seeing it ourselves that we can go from idea or intent to trying something, experimenting something, getting feedback, iterating on that, and then being able to firm it up and deliver to customers and again learn along the way. So that's something that's happened very clearly for us and that shows up in even products. When I joined, we launched Cortex code, but that product was done really fast because that loop was possible. Right. We were able to do something prototyping, we got a lot of internal users adopted. Then we went to customers, got feedback and we were able to launch, which you and I would not have been able to launch in a few months in a real enterprise product, which is possible today. So I'm pretty stoked about that opportunity right now more than anything else. And the second thing is, of course the way I think about this is we really care deeply about at Snowflake and what we do in terms of using that experience to say how do we make AI real for enterprises? Right. That is kind of the core thesis of what we're trying to do every day.

Speaker A: So let's focus on the latter a little. And selfishly, I'm excited about that because what I do day to day is talk to enterprises and this is top of mind for everybody. Like it's probably number one initiative. Just curious, as you talk about making AI work for enterprises, the ones that have been successful, I just think the gap is growing so quickly between folks that are successful and the ones that are not. What are some of the commonalities, the two or three things that you've seen is common across the ones that are successfully doing it.

Speaker B: So I think when we talk to a lot of customers, I think starts in three buckets. The number one bucket is I think there has to be trust that whatever they're trying to do with AI will work and is very clear in terms of what is possible, what's not possible, what is experimentation and what's non experimentation. Right. So to us, I think it all starts with trust. And I think we have a very clear advantage specifically there because we've built this company over the last 10 years on customers trust because we handle sensitive data all day long. So that kind of comes naturally to us. And that's kind of the foundation. I think it's going to be super hard for enterprises to make anything work. Right. Number one. Number two is I think we still have to deliver value and solve for really critical problems that the enterprise users and their customers are Facing the way I at least try to encapsulate that is the value is the new virality. If you can deliver value, I think then you can get viral. Otherwise there's no virality in this day and age. So that's the second thing I think about. And the third thing is really about reducing the time. I think every enterprise is under this pressure to deliver value to their customers. We should think about AI with the trust and, and what problems you're solving to see can we go faster? And when I say faster, it's not blind speed, but it's the ability to understand, learn faster and get to the kind of end state and then keep going from there. I think that is the third element. I think if enterprises who do those three things well and companies like Snowflake who do those three things well with the customers I think eventually succeed. The one other uh, thing I would add to that is what it also is interesting now is I think there are a lot of things, I'm sure you've seen this in enterprises for a long time people put aside because it is too hard a problem to solve. I think you can start to open up and take that out of the shelves and say okay, can we solve this now? Can we solve this now? I think that is equally possible, which was not possible earlier. So I think the core three pillars continue to be the trust, solve the hard problems and deliver value. And number three, can you move at speed and learn faster. But the other thing I would leave every with is that now you actually can be more ambitious than ever because the cost to go try something which is hard and impossible earlier has come

Speaker A: down a lot, uh, a hundred percent. And the feedback loop is so much quicker. And I would say this is the advantage, I would say startups have, right? Like you don't have to change management, you don't have some of these downsides or risks or the political things that later stage companies have. The I'm curious that actually you brought up something that I think fascinates that uh, everybody talks about. And you have such a unique perspective cause you've been on the engineering side, you've been on the product side, you've been on the customer facing side which is time to value. And I don't think you can provide the whole project as a value right like right away, but you can at least show some quick wins that get you points and show people that we're moving in the right direction.

Speaker B: So let me start with a bit of a behavioral thing and then I'll answer This question behaviorally, I think we think of ourselves as you got to have a mindset of a small, nimble startup company, like it doesn't matter how big you are. And that, uh, kind of expectation we have within the company, that means we no longer as a product review, we don't do things like mocks and PRDs. Show up with the working prototype so we can actually learn from it, critique it and make it better. And we can then take it to the customers and also have that feedback loop.

Speaker A: Right.

Speaker B: So that's the true way to think about it going forward. And then we can even go one level further and take that and work with the customer, uh, in whatever FD ish model to actually make it into something real for customers. Right. So that's a behavioral aspect of it. Going back to your trust and quality statement, I do think there is. The good thing is that we are able to reduce a lot of the time. That used to take a long time. So, for example, we are in the data business and people, as you've been in the data business, migration used to take a long time because it's a project you need to plan. People are involved. I think with what we're seeing with tools like Cortex code is we have skills which allow people to make that migration go faster, do the quality checks and make sure that things are really set correctly before you do any kind of other tasks on top of that. So that is actually very helpful with AI, which is not possible earlier. The second thing is, I think once you have the data and the governance and the context there, then I think what happens is you are able to now use AI and that actually context becomes much more of a useful thing for whatever you're trying to do. And that is a flywheel you create, which is that you have the data that's governed. The AI is able to do something with that and then it builds on itself and the speed picks up. It's almost like a flywheel there. And I think this is where we are finding ourselves in a very interesting spot. That because of that foundation of trust we have and the ability we have with the Cortex core kind of tools, we are now becoming much more of control plane where people are able to deploy this without worrying about both the speed and the governance and the quality piece. Right. Because now there can be thousand experiments that can bloom and 100 will survive. And that's fantastic. That's what we want. So that's what I think we are seeing in terms of how people are

Speaker A: balancing quality, is sometimes customers just Move really slowly. And this is what I struggle with. Where some customers are still happy with their on premise software. So I'm just curious, how have you tried doing that? Let's call it change, uh, management, challenge management. Any tips there?

Speaker B: I think it's a very valid question. We have customers like fanatics who are able to use Cortex code to move things in like in a weeks and hours which took months. Right. And the same thing with a large customer like united rentals, over 1600 locations, being able to make data accessible with natural language for their folks employees without having analysts who have to get in between and help them out. And the reason I'm giving these examples is the best way to actually get the confidence of a customer to move fast is, is to show them, yeah, it'll work, right? Like I can, like gone are the days where we could go with a slide deck and a timeline and saying we can actually do this in a number of months versus whatever. Like people like, yeah, I've seen too many pictures, like I don't care, right. But when you actually are able to do that, or sometimes even they are able to do that because we provide them the tools and skills, tools like Cortex code, they actually get the confidence they can move faster. So that's the number one thing that has changed for us is people's ability to actually see the evidence that they couldn't see earlier. And now we also have partners, for example, who block saying, hey, I went into this customer, we started this migration project and it closed in two weeks. And can you imagine, like partners, you've so many partners in your life, none of these migrations are possible that fast. Right. I don't want to call it formal because it sounds um, negative, but it actually creates a positive cycle of everybody saying it's possible now let me do it. And that's how I think we unlock the speed for enterprises. And then last one I think is, I think every enterprise is worried about how AI is going to disrupt that business and that means they have to find ways of moving faster. I think that's where we can be very helpful and we try to take that approach saying, hey, what are your business problems? How can we solve that and how can we help you go faster with the tools we have? I think that ultimately helps out these customers.

Speaker A: Yeah, a hundred percent. I think you've nailed it across the board. One of the questions I've been really interested to ask you ever since this was scheduled because again you have such a unique background is time to value, time to build, is Diminishing, Right. It's becoming a little more easy to compete on. Like the technical differentiators. If the moat is no longer product or less product, whatever you want to call it, what is the new differentiator like, how do you build a more defensible company, split it into kind, uh,

Speaker B: of behavioral values and then what's the execution? Right. I think behavioral is to me, the clarity of thought. I think we have to really bring the organization aligned on what are the problems we are trying to solve, for whom and how are we going to execute fast like that? Clarity has to exist for differentiation. The second thing is I think we, even though I told you the company has been founded 10 years ago and we continue to build more and more interesting innovations, it's still grounded on core values of being easy, connected and trusted. And if you think about it for a minute, and, um, those things matter no matter what going forward with AI, Right. So that's the second piece. But the third thing, which I really like a lot, is that I think there are still a lot of problems, really hard, persistent problems enterprises have. I think if you stay focused on them, I think the differentiation will sort itself, uh, out. Sometimes you need to really lean into the hard, persistent problem that has not been solved is what you need to solve. And that differentiates you versus trying to say that, hey, models are all the same, everybody has the same access, everybody has a time advantage to value. I think those things do at some point pan out and you have to really focus on do you understand the problem deeply and can you solve it for the customer and drive the value. And you've done this before in with past years, really crystallizing that value for the customer.

Speaker A: Doing it in practice is obviously a little harder. But I do think you mentioned something that, which is echo, which is the more you can learn about just being able to talk their language makes such a big advantage. I was on a call this morning where it was supposed to be a pricing call, but I started with a sheet of all the things we've done through the project. And the customer even called out, they're like, oh my God, you guys have done a lot for us. So it's sometimes just indicating that you've gone the extra mile. My last question, what gets you excited moving forward? I'm asking that question is, at least for me lately a, uh, lot of people with the adoption of AI and whatnot have been a little nervous, a little scared, and all those feelings are reasonable. Valid change is not easy. But I'm trying to showcase the other side, which is, hey, there's a lot of exciting things happening as well.

Speaker B: First of all, plus one on what you said, right, I think we have not seen this much of change compressed, um, in this kind of shorter time frame. We've all been including you have seen a bunch of cycles, technology cycles. This is unlike any. So all those feelings are I think very natural and I think we should not ignore them. The three things that stand out to me are I think the speed of learning and decision making is I think changing at a speed which never I think the number one thing so you can go from intent idea questions to answers learning very fast, which is very important. And the reason it's important is I think where we in the B2B space, we've talked a lot about data driven organizations for such a long time, but now I think it's the reality where actually it can happen because we can bring data and the natural ability of AI to bring out what's needed for, for each user to make their lives data driven. And to me that's important because it actually empowers people in an organization and it empowers them to take better decisions and that flex back on the organization and hence it's better outcomes for the organization. So I think this combination of AI and data, uh, empowering people and truly driving data to an organization I think is the second thing I'm super excited about. The third thing is I think we are already seeing that over 9,000 of our customers are using our AI AI capabilities or 50% of our customers are using context code. And the reason I give you those stats is I think we democratized AI to everybody, whether it's technical, not technical, whatever your role is. What's exciting for me in that is that not everybody is a builder. They can build what they need for them to be successful, for them to be impactful. Right? It's no longer I have to wait in a queue to talk to somebody to help them and then they help me to get somewhere. I have the empowerment and the ability to do that. And especially in my role, that's super exciting for me because our team is all about how do we help builders and we don't want to differentiate anybody who has the interest, who wants to build. We want to help them and make our platform the best for them.

Speaker A: Right?

Speaker B: So those three things I think are the most exciting things for me on what the future holds. And we are very much keen to looking forward to helping each of those Personas have their best impact as we

Speaker A: go through this transformation plus one across the board, the barrier to entry so low and it doesn't have to be snowflake related like the whole ecosystem with some of these, let's call it vibe coding products or whatever it may be, has just made it so much easier for anybody to build what they want to build their vision. Bala, this has been fantastic. I just love how crystal clear you are on your responses. The one question we love to ask before we say our goodbyes is if Bala could go back to any time and I'll let you pick the time. The wiser, older. Bala, what's one piece of advice you would give your younger self?

Speaker B: Uh, coming from India as an immigrant, right, I grew up thinking about you got to study, you got to prepare, and you got to be right to have impact and have a career. I think if I can go back, I think the advice I would give to my younger self would be what's more important is to learn continuously and grow continuously and be very curious. I've, um, taken a lot of the pressure off myself and, and we're in a much more interesting place mentally and what we could do for the world. If you could take question.

Speaker A: It's so funny you bring this up. I've told myself, look at life as a journey and not a destination. But it's so easy to lose track of that because we're always like looking for the next promotion or that next outcome that we forget, hey, enjoy the rides, enjoy the waves and take the good and the bad. So a great reminder. So, Bala, this has been fantastic. I can't thank you and your team for coming on, for supporting us, for paying it forward. For anyone that wants to listen, I, uh, will put Bala's LinkedIn and whatnot in the show notes, reach out to him, thank him for coming on the show, and until next time, be well, be safe.