AI and Customer-Centric Leadership – Min Wang, Splunk

This is Textron tv. Hi. Really nice to meet you.

Nice to meet you. And, uh, this is Been Wing, by the way, who's cto? New cto, relatively Speaking.

Yeah. Getting, getting all that news very soon. And, uh, I joined the 91 days ago from Google, and it has been a fantastic journey so far.

Not like a fourth joy, no challenge, but nevertheless, that's a fun is about really. I will talk a little bit about, uh, my own journey and, uh, I did my major in China HAI University Computer Science. And, uh, I got my master in computer science as well.

And then I came to us, got my PhD in computer science from Duke University. Mm-hmm. And, uh, I spent, uh, the good part of the first 10 years of my career with IBM TG Watson Research and invested in New York focusing on data management and information analytics in particular.

Actually, I always call myself a data person because the first thing here was about relational database. I actually work on DB two s query optimizer quite a bit. Very interesting.

Yeah. And then I left, uh, IBM after 10 years, became the director of HP Lab China and was on international then segment for three years. And that was a time I started to extended my horizon beyond database.

Mm-hmm. And then I joined Google first time when Google Research was still quite small in Mountain View. And then I left Google, joined the visa, the financial, the payment industry to start the Visa's, uh, research organization from scratch, and then went back to Google, all leads drop up to that point.

That was 2018, you know, 20 years. I always lead research team, research organization. Mm.

Industry research. Uh, uh, that was the year in 2018. I made a call saying r and d.

I always add the R side. I love it, but I want to boom. It's actually quite unusual if, you know, the tech industry people stayed in r if they do transition, they usually do muchly.

Five years is a sweet spot. You tend to be a researcher or a product or practitioner. Yeah.

So, so I did the transition in 20, after 20 years I being in the research side, because over my career, like in the research side, the more I did it, the more I feel like my passion is moving to really touch the end, the user of the customer. Mm-hmm. And I don't want to go through, uh, like R two D transition to another team to deliver the value.

Right. So that's why I came back to Google, actually started to work on Google or build a new product, and then did that two years, right at the beginning of the pandemic, I moved to Google assistant, which really focus on build a digital assistant for consumer using ai. Right.

Great. Great transition step to Come here. Yeah.

So the reason I came here was really with really the advance of generative AI and see how much energy and new effort it really happening at Google around that area mm-hmm. Made me really to think this power technology. I wanted to be applied to a critical industry and help enterprise in the things they care the most.

Plunk is really the place. Mm-hmm. There's a lot of great connections in your background being a data person.

By the way, I used to be a DV two DBA a long, long time ago. So talked many years ago, but when it was relatively new too. Um, but, but of course, you know, being such a, a data management platform as really Yeah.

And part of what Splunk is and of course your interest in ai. Was there a security aspect of your background? Cause so much of Splunk's customers are using Splunk for security or no more data kind of AI has has been your, Uh, I, I have actually, I'm learning security if I have to say I need closer in the neighborhood when I did my master in computer science in back in China in Shanghai University.

My dissertation is on cryptography. Oh, okay. Perfect.

Yeah. What a great, what a great kind of fundamentals of security, right? Yeah.

Well, tell, tell. So I was, I was really jealous yesterday, you know, in a good way of you being able to stand up on stage 90 days into the job. Talk about the AI announcements, the, the preview of the, uh, the assistant, the Splunk assistant.

Um, as you walk in into a new role in, in a well-established, well-known, you know, liked company, what were you kinda looking at to try to assess, there's always the interview and then there's the, the position when you get the job. Mm-hmm. What things were you looking for to try to understand what's happening at Splunk or with Splunk customers or the technical side?

I think there are, there are, uh, a couple of layers, uh, with your question. The first layer was easy. One is really, uh, my interview and, uh, when I decided to accept the offer, there is a really, what I really wanted to do versus the role the company want me to do.

Mm-hmm. Right? And that one, fortunately is a hundred percent overlap.

What I want to do, I want to use technology and innovation to deliver best value to our customers. Really, you, you, you, you, you, you probably notice a phrase is actually very clear. Technology and innovation is a means plan.

Mm-hmm. The end is customer value and the, and the, and the company is heading exactly to that direction. Right.

Then you come to the next level, how? Right. And that is a part I'm still working on.

Right. So I, I really like, as passionate as I am with AI and generative ai, I'm a firm believer is really, you need to identify your goal for your customer first, and then find the best innovation technology being all new, all existing to solve the problem in the, in the most effective way. Mm-hmm.

So I really want to call, you know, as powerful as generative ai, in a lot of cases, it's not the best technology for solving certain customer problems. Mm-hmm. And it is not suitable for certain situation.

And the traditional database statistics, predictive analytics, classing, are still quite powerful and they may suit for your problem. Right. Better, and they are better proved.

And, uh, generative AI is powerful Yeah. Beyond everything else we had in the past. But it is also have a lot of, uh, challenges like the, the complexity of the model, the probably nature of the, of, of the whole technology, of course hallucination.

Mm-hmm. There's, unless there is some really revolutional breakthrough with how we build a large model, there is no way we can eliminate hallucination. That's, uh, I think a lot of folks are trying to understand about generative ai.

What does it take to train models? What does it take to build models? Kind of two different path, but it connected and of course, prompt engineering, there's a lot of new things.

Yeah. One, one of the things really I think really important about what you mentioned is that's partly that transition from research. Yeah.

Cause I've worked in research myself too, from pursuing what can we do and, and advancing the technology to, it's about outcomes. It's, it's making an impact, um, that, that people will pay for also in a, in a, in a for-profit company. And that's, that's part of that transition.

And there's a nice pleasure to, that you get a, a real sense of satisfaction. Yeah. You see a commercial on, on TV or up on a billboard, or you talk to a customer that's using your technology.

That's always been a big motivator for me. How about you? Exactly the same.

I, I really remembered, uh, when I got my PhD, uh, I had like a two passive potential. One is really, I never thought about, uh, like really, uh, do product development directly at that time. Mm-hmm.

You know, I thought either became a university faculty or became a, a researcher, but, you know, really applied the well established industrial setting. And, uh, so I interviewed with Bell Lab, uh, uh, IBM research and so on. And what made me feel is really I want to be closer to real impact.

And, uh, and this is like, uh, always drive me get closer to, closer to the more realistic part of r and b, to identify the value for the users, end user of the customer. Enterprise customers. Yeah.

What's the, what's the, uh, the thing that maybe you didn't expect to learn about Splunk that you've already learned in, in the short time of 91 days? I think there are couple things. One is really, uh, I think I really love the culture and, uh, I'm always a very open, upfront person.

And also I have a, like a very critical attitude. Like I want to spot what can be better. Mm-hmm.

Uh, instead of, uh, put a lot of energy to energize people on the positives. Like, uh, my kids know that making Things better. Improvements.

Yeah. Cause I care so much. So I, uh, for me, that always translate to, I want to, like, I care so much.

I trust we can do better. So this is like something we could do better, right? Mm-hmm.

I think, uh, this like, sometimes in certain culture doesn't go really well. Mm-hmm. But here I just feel like, uh, at least the people understand where I come from and, uh, I could be myself, uh, like either being asking stupid questions that sounds maybe like I'm doubting people, but even I don't, I just want to learn or saying things like, why couldn't we do it faster and push really hard?

I think this part, I really enjoy being myself really. Mm-hmm. Like, uh, this is not a place I need to put a filter to talk about technology or talk about the organization.

Right. And the second part is really, I do feel our company is in a stage of enter the next level, but being more focused, more focused, like really deliver our key values and drive one the main initiatives in this industry. There are so many challenges and opportunities.

It's easy to do things that you justify is useful mm-hmm. And helpful in certain sense. But we need to be super clear about our critical mission and what are the expect outcome and more important thing, the measurement about our progress and how people come.

It's that focus, right. There's too many things you could do. Yeah.

Right. What are the right, what are the best things to investing time in? And, and what I'm sensing is a real curiousness part of that, making things better is and asking those dumb questions.

Yeah. Sometimes maybe they feel challenging and they're not meant to be. It's really that curiousness, kind of like Steve Jobs document.

Be curious, but that's where innovation comes from Too. Yeah. There's several questions I asked my team in my first couple weeks.

I said, tell me what you are working on without using any jargon. Because I'm new, I don't have thisAnd all, all the acronym name. Mm-hmm.

And, but I'm a, I'm not dumb. I have common sense. If you are doing something, explain your way without driving so I can get it.

Then second, tell me what customer value you bring this work. Okay. Third, how do you measure the validate the customer value?

You just told me, right? Mm-hmm. So these are the easy questions and uh, actually this is not originated from me.

Uh, one thing helped me a lot, uh, in my career is, uh, I don't know if you know this person, George, help me. Who use, I don't, uh, who use, uh, who is a famous scientist used to run DARPA in 1970s, okay? Mm-hmm.

Uh, go search Google help catechism. He had a infamous famous list of question. When people submit proposal to darpa, he ask, and I ask myself and my team all the time.

First is really just, I said, tell me what you're trying to do without interrupting. And second, who cares? Uh, how we measure your progress.

What is your, uh, meter? How much is cost? What is end result?

Mm-hmm. What he's done today. Mm-hmm.

Right? So there's at list of fantastic simple questions. And if we go back to that list all the time, I really feel like a lot of people will feel, oh my God, I probably need to think about what I'm working on.

Because there is no measurement, there's no meter exam. The success is not well defined. All success is well defined, but does it matter?

Mm-hmm. Um, so, so I know you're learning a lot about Splunk, right? You don't know everything in 90 days.

You're curious, you're always learning. com and some of the announcements that have happened, of course, you talked about AI and generative AI as your talk on the main stage. Um, are there any things that kind of stand out, like I'm really excited about, obviously AI have important to you that are any other areas in the, in the new announcements that stand?

I mean, they're all good. They're all important. Yeah.

Any, any sort of bubble up? I think, uh, there are two things. One is really the concept of, uh, unification.

Mm-hmm. Really like, uh, I think one biggest challenge for enterprise is really data, uh, isolated or process are isolated. I, I think Splunk is in a powerful politicization bring platform, security observability in one place.

So you have full visibility about what's going on in your company, uh, inside of like their solutions here and there. And they don't talk to each other. I think we still have a long way to go, but we have a good foundation.

I'm really excited about that. The second part is really when it's come to ai, I think setting up a strategy, ores one thing, put it into execution. I really deliver my value to customer is really my next major focus in the AI area.

And the current thinking is really we, we need to strength the team. You know, uh, in my view, I need success in using AI requires everything. One is clear defined business expectation.

We have it here, right? Help our customer to better build, uh, to resilience, right? So that's outcome, but then we need to translate it to measurement, right?

Mm-hmm. And I actually just, uh, during the past couple days in Vegas, I start to say like, I could tell Carry Town and everybody else, like using our goal and fact outcome is use AI to help us are, enhance their capability in building digital receiving, but then the next cycle, how I can measure you. Mm-hmm.

Uh, my current thinking is really is can be around several dimensions. This is like, you probably is among the first to here that like, it is solve scoop here. You got, you got it here.

Yeah. It's, it's really like, I think it's really in their common workflows mm-hmm. On the day to day job, how we, can you say other technology to make it easier, faster, simpler, smarter.

Mm-hmm. You notice I didn't put cheaper here. Mm.

And the reason I don't put cheaper, actually this maybe only me, I think cheap, cheap or cheap or expensive is, uh, not well defined term. It's all compared to the value we buy things. We don't look at the AppSec would price to see if it's cheap or expensive.

We are actually seeing cheap or, uh, expensive based on the value we perceived mm-hmm. Or received. Right.

This could be different, but I, I really think I would say like focus on the value and the matters of progress around this four dimension. Maybe there will be five simple and, uh, easier. Maybe have some overlap.

I'm just at the beginning of my thinking about different metrics. Mm-hmm. So different metrics, meaning like, we identify some key use cases, and then we start to say like, like I Google, there was one product I did that I literally really measure, like the faster part is really the old workflow.

We got interview a lot of like users, so we use it, right? How long it is, take them from this button to that button to get one particular thing then, and then with the new flow and the new in the backend, how much faster we got. Right?

So this is one way to see the acceleration of the process. Right. Smarter is a tough run.

Right. And, but we got to have some way to measure it. Mm-hmm.

Right? I think smarter related to the gent, both quality and the brace of the capability. Right.

So this is one thing. The second thing is really start to identify some common building blocks. Yeah.

And, uh, because well, everybody's needs could be different and there, there are a lot of commonalities. After all, we all try to protect our digital systems. Right.

And, uh, uh, my current really like God's feeling, like gut feeling usually is a dangerous thing because most of them are wrong. You need to find some data to validate it or Test. Correct.

Yes. I, I really feel like start with, uh, uh, summarization, uh, on the question that answer in the domain specific mm-hmm. Report knowledge could be a good start.

Meaning, because obviously I need, they spend a lot of time to read, shoot reports, read this and that. And, uh, AI in particular actually could be quite powerful in doing summarization. Right.

And, and then you, instead of reading a report for two hours, you actually got a summary that you can read in 10 minutes and then you ask five questions. Mm-hmm. Right?

And it, then you save your time for you to collaborate and do more, uh, higher level tasks. Right. That Actually, it, it that fits well with one of the things that I hear from, from practitioners companies all the time is we're all being asked to go faster.

Yeah. Sometimes we're being asked to save, save money at the same time. Yeah.

Um, but it's being more competitive, being more agile, you know, technical teams are being asked to do more mm-hmm. With less, faster, et cetera. Yeah.

But at the same time, it's getting more complex. We're moving more to the cloud or adopting cloud native or whatever things that we're doing. Yeah.

Um, and, and security's a tough one to hire for already. So it seems like one of the areas we can help people with, you mentioned workflow mm-hmm. Is just making the job easier.

Right? Yeah. So your summary idea mm-hmm.

You talked about is all about saving the workflow, connecting all the dots, adding the context Yeah. To it. So here's some answers, selecting that now what you might do in your workflow.

Yeah, yeah, exactly. Exactly. So, so a lot of things could, could be streamlined and uh, also reduce redundancy, right?

Mm-hmm. How were you, how were you, um, so maybe a year from now or so, how will you assess kind of how well did I do in my first year at Splunk? What, what things might you look at?

I know it's hard to predict the future, But Yeah. I, I, I think about that all the time myself because I, I, I feel, uh, number one, what will make me happy a year from now is really some of the key use cases we identified, validated with customer. We already start to build certain capability at a production level, not a demo.

Okay? Mm-hmm. Not a demo.

Mm-hmm. Like it, this is a huge gap between a demo and uh Oh, absolutely. Into the product, right?

So, so I think that's one thing will make me super happy and, uh, uh, it is going to be hard. I'm challenge, I know because, uh, uh, the technology have is own challenge in terms of dealing with hallucination, uh, and so on. On the other side, there's a challenge about integration mm-hmm.

Because our product is huge and complicated. Every team have their own roadmap. How you really pull everybody together so that we work on a common workflow instead of building a AI app as a standalone.

That's a last thing I want to have. Mm-hmm. Okay.

So the second one is really for me to became a Splunk CTO in instead of a Splunk ai CTO right now is I feel like I'm Splunk AI cto. Right. So to really have a much better understanding mm-hmm.

Of the tech stack product, the solutions, and also how our team work together. Mm-hmm. So I interact with a lot of people during the nineties days, but I would say 80% of them because I want to talk with them about ai.

Right. So, but Splunk is not built on top of AI 20 years ago. And AI is going to be a big new portfolio in Splunk, but is not going to be all.

And it took, it will take us a while to really like be helping our customer and we have to understand how AI can really work together with the rest of Splunk. I would imagine most of your customers are in that same journey too, right? Yeah.

They're not built on AI for 20 years. Yeah. They're trying to assess Yes.

Yeah. Maybe some of that learning too on that Journey. Yeah, exactly.

Yeah. Well, fascinating. Thank you Min, for spending some time with us.

I like chance this topic and uh, and uh, okay. So the next thing is really if we meet here, here, uh, call me out here. If I still only talk about, yeah.

Okay. So we'll sit down. I know it'll be in this room, but we'll sit down at the next uh, dot com.

Yeah. And that's, so kinda check in. Hopefully we get to talk before then again.

Yeah. But yeah, we'll kind of, you know, let's check in and see what you were thinking then and how you're thinking a year from now. So I wish you all the best and exciting, I'm sure for the Splunk team to have you in this role.

So, Welcome, the team is fantastic, and, uh, this is, uh, like a big part of my happiness. Like, you have to feel happy around the people around you. We spend so much time at work, right?

Mm-hmm. There's been some great announcements, um, from Splunk around ai. com.

There's other announcements, um, but I would follow 'em in and, uh, as someone who you can kind of look to and learn from her while she's learning and of course share with her. So we hope we get a chance to do more of that with you. I'm actually looking forward to letting all from all you.

Yeah. Thank You. Great.

Thanks Ben. Mm-hmm.