Real-Time Data and AI Innovation with Hasura’s Suku Krishnaraj

This is Techstrong tv. Hey everyone, welcome back here to Techstrong tv. I'm really happy to introduce our next guest to you.

It's the first time he's been on here with me, but as we were talking off camera, we, you know, six degrees of separation in the tech world. Um, I want to introduce you to Suku. Suku.

Krish Niraj. Krish, Niraj. I think I've said it right, Suku, is that right?

You are totally right. Yes. That was great.

Okay, Thank you. Suko is the COO, chief Operating Officer and President for go to market at hra. Um, we're gonna find out more about her and, and everything in just a moment, but let's find out a little bit more about Suko.

You know, we were talking off camera. We, as I said, six degrees of separation in the tech world, but Suko give people a little sense of your journey of how you came to B-C-C-O-O and president here. Awesome.

Yeah, absolutely. Uh, first of all, thrilled to be there, Alan. Thanks for having me.

Um, my pleasure. Uh, I've been in the tech industry for over 30 years now. Um, the journey is a long one, but let me, let me see if I can summarize it.

Um, I began like career as a developer, very technical. I was building large scale distributed systems, um, you know, when payment software for the first six years, uh, moved to the states, and, uh, and I moved, you know, I was not, um, I was not particularly, uh, enthused with sitting and just coding. I was a people's person wanting to interact with customers and et cetera.

That's how my product journey began. Uh, I started, uh, going meeting with customers, sales and whatnot. Uh, did product management, product strategy for over 10 years, uh, sort of the data section of developers and observability play, um, for, for mid-market and large enterprises.

Um, then, uh, I was with this company called SolarWinds. Well, during my product journey was with Mercury and Ulet Packer, and then I ended up in SolarWinds. I really, I Know them all.

Yeah, you know, you know them. So it was phenomenal. Um, up until then, my experience was like going top down, basically, um, talking to customers and selling to customers in Prises, but SolarWinds taught me something different, which is how do you actually flip your funnel upside down, go to a wide mass of practitioners and still grow very fast and scale the company?

And, uh, solar would taught me a lot about data-driven marketing, data-driven, uh, go to market. And, um, I was there. And, uh, from there I went to, um, CenturyLink.

I was a GM for almost close to a billion dollar business. All of their cloud offerings, uh, very different, right? Uh, having a very modern SaaS company experience to send Link.

Um, it was, it was great. I learned a lot, uh, as a GM learned p and l. Um, from there I went to Sumo Logic.

I joined Sumo. Uh, you know, it's in the data analytics space and, and lots of metrics traces, um, providing ity and sim. Uh, I joined Sumer when it was 10 million in revenue.

Uh, we went through multiple rounds of funding. We went through, we talked a little bit about going public. We went public during covid time, subsequently a private equity exit.

Um, and when I exited, it was 3, 3 23 30 million revenue. So phenomenal experience. And then, um, I, I wanted to take a break.

I was on a sabbatical. Um, it was supposed to be longer, but, but three months, but it never is. It never is.

Uh, I met with, um, through the Lightspeed guys. I got introduced to Hanai, the CEO, um, and, uh, he's a brilliant, brilliant, uh, founder, um, of asura. And, uh, we just got talking and, and basically he was, he was looking for a business partner, if you will, somebody that runs the business side, the go to market and the operations.

Um, and for me, um, it sounded perfect. I could, I could take all my product driven engineer product, go to market experience and apply there. And, uh, we are barely scratching the surface in this space.

And it was kind of, uh, match made in heaven. And I, it's been close to four months now, Adam, and I'm loving every day. So that's my good for you, man.

In short, short stint, that's my journey. The growth scale guy, very technical, um, developers, SaaS, infrastructure oriented background. You know what it, it, I can't say I'm not familiar with it.

It, it's, uh, the path and stuff that I, I know. Well, you know what? I think our audience that we, you know, our audience is, is a tech audience, but it comes from different areas, from developers, from cloud native ops folks and platform engineers and, and steber folk and, and all of these.

And then of course, now everybody's an AI expert, right? Um, some of them have heard of er, but I bet you a lot of our folks have not, or if they have heard of aa, they're not really sure what, what they do, what ER does, what you guys are about. Soko, if you wouldn't mind, for those who don't know her sir at all, or think they, or they've heard, but they're not sure, how would you describe URA to them?

Yeah, absolutely. Let me, uh, I'll take it in three parts. One, kind of really describe ura, what we do, um, and give you a sense of a little bit of a journey of the company and then the future of the AI piece.

Let's talk about three. We are in a nutshell, uh, we simplify data access for your developers. Um, we bring data in a very simple, unified way that developers can use regardless of where the data is located near enterprise.

It could be in, uh, sql, it could be in no sql, it could be in, uh, SaaS applications, API to some legacy systems. We provide a metadata layer that, um, with a, with a unified schema that allows you to go and get the data that you need in a very instant fashion. This is the big challenge for enterprises today.

The problem they're facing is any development of any new applications, and they need teams of, uh, developers. They're spending wasting time, uh, going surfing through all of the data that exists, spread out in, in the, in your enterprise. And, and a lot of the other vendors come in and say, you know what?

Bring all that data together or make it simple and easy for you to access it. We say, no, keep the data where it makes sense for you, Mr. Customer as a business.

Leave it where it is. But we, you can unify that through a metadata layer. We will help you build a super graph that allows you to join the different data sets and then simply A-A-A-P-I connection with internal and external applications.

So the business benefit is, is we help you get to market faster. The new applications we help, uh, access to the data in a secure, reliable fashion. We help you maintain the data governance, uh, quality consistency, which is a big issue in financial companies as new business units are getting spawned up, and teams are using it in a, in a very non-standard way.

And it's a problem for centralized architecture groups. So that's the business value. And we sell to two discreet sets.

One is stronger than the other, one of the CTO organization where they really care about accelerating their development of apps. The other one is the CIO organization where the data analytics spaces, right? Um, Alan, we are fortunate, uh, also one of the reasons that brought me here.

We have some incredible logos that are entrusting us with this mission critical layer. Apple, Cisco, JP Morgan, chase, bank of America, Optum Health. So, um, phenomenal logos.

Now, if you think about it, let me take that to where this is headed. Um, just if you think about ai, right? Which company isn't trying to build a AI assistant today?

Every company, everyone, everyone. Um, and, and the biggest challenge today that I'm hearing from customers is that, um, you know, LLM, what if I could apply LLM to the enterprise data? What if I'm able to answer questions that are meaningful today where I have to bring in a team of data scientists to get my questions answered, whether I'm a CMO or a CEO or chief customer officer, uh, being able to ask support question, right?

Um, you look at Einstein and you look at other existing, um, uh, AI platforms out there, there's a lot of hallucination. It doesn't connect to the enterprise data. Take what I mentioned earlier about unified data access layer to ai.

If you're building an AI assistant today, you need data, and you need, you need the access to data in a very simple fashion. io/prompt ql, um, it, it provides the, I call this the magic in a magical fashion. You bring your own LLM, you connect from QL to all of your enterprise data.

We help you answer the questions, answer questions like, um, who am I top five customers that are, that are at high risk? We combine just the query planning and just showing the results from this data with computational capabilities. For example, if you want to know the high risk customer, we actually prompt QL gives you a risk score.

It actually tells you how it is coming up with this risk score. What are the attributes it's using, what tables it's accessing to give you the risk score, and it sorts, it reverse order top five customers, high risk score. And it tells you exactly what it is.

Now, you could either, uh, programmatically have it go deeper or you intervene as a user and use the LLM to interact with it. Um, Alan, we are seeing tremendous, um, pull both from existing customers as well as new. Uh, we were a dream.

Um, and I don't know if you were there last week. Um, we, we got, yeah, so I was there. Yeah, you were, you were there.

We missed each other. But, um, folks that stopped by, it's really an enterprise. Uh, you ask them, uh, are you trying to build an AI assistant?

It's like, yeah. What is your biggest point pain point? Well, I need to be able to trust it.

I don't have a trustworthy AI assistant today, and I have billions of dollars running through it, but I can't trust it. Um, why? Well, enterprise data is spread out everywhere.

Uh, there's hallucination. The existing LLMs and AI assistant don't give me accurate information. So that's the problem we're solving.

So I'm the vision for the company, uh, at the back of this data layer and the API layer, we are building an AI on top, and, uh, it is truly revolutionary for us. And, uh, and I'm very excited to scale soda with, at the back of this AI value proposition. Sure.

So I, I was at AWS unfortunately, I spent 90% of my time in our studio at the Winn. We had one of the suite power suite rooms where we were doing video there all week. So I wasn't on the floor as much, but I I didn't miss when we got on that floor either.

It was a bit crazy. Yeah. Um, but really we, you know, Suku, we're, we're talking about two things here that I think are important.

Number one, I think organizations are coming to understand that to really milk the benefit of AI for their organization, you need, you need, whether it's custom LLMs or you using a rag, you know, type of of situation, you, you need that customized dataset. Yeah. Right.

Pretty much that it's pulling from. And, and I think a, a choice coming down the pike for many organizations are, do they, is it worthwhile for them to have custom data that custom LLM in the current state of what it takes to, to create one? And if a server could make that easier, right?

To allow everyone to have in essence, a custom LL lab, whether they're using RAG or whoever, you know, what's going on under the, it doesn't matter. We don't have to get into, right? Yeah, yeah.

But If, If I can make that AI work for my unique data set, not just in general, now rubber meets the road there, right? That's where things get real interesting. That's number one.

Number two, again, the, the trust issue and, and there is the halluc poisoning piece of it, and I, I think we're at each iteration of, of ai, we're, we're getting better at that. But there's another piece of that trust issue, which is that AI and the answers the AI gives you are only as good as the data it's tied into, right? I, I, I'll give you an example.

I, I went on, I went online last night for a popular, uh, shoe. Uh, I was buying my son a, a pair of shoes for holiday, and I forgot I had it shipped to my address instead of his address. So I went in like immediately I realized that I went to go change it.

It said, use our AI empowered chat system. And I, I went in and I, what's the problem? I told them, I wrote everything down.

It asked me three times. It was very light, it was very responsive, it was very everything except it didn't have the ability to tie into their ordering system to update the address. And so it said, okay, we've got it all.

We'll have someone contact you in 24 to 48 hours. In the meantime, today, I got a notice that the shoes were shipped out to my address, not my son's address today. Oh my God.

Yeah. So, You know what, so they, this company went through all of this to get their AI stuff straight, right? To have a, an AI chat system where it just, it wasn't connected.

It's not connected, it wasn't, It's just a pretty, and not only that's a pretty face. Yeah. Not only that, it's the, it's the action.

You're basically asking it to date the shipping address and please change the workflow. Um, so that's the unique thing for us. I mean, you talk about RAG and others, the way we differentiate, not only the fact that we query, we give you a very contextual query, but you can attach actions to it.

Hey, go update the order. If this is the case, then, and you as, as a user can tell it to, or I could automate that through the metadata and one of the nodes, the invocation is the action. So I could even invoke another LLM if I have to.

So check it out. I, you, I think you summarized it really well, which is, you know, the trust is a big factor. And then enterprise data for LLM is, is, I think it's a game changer.

Uh, and customers don't even Know what Oh, I, I think, right, you, you're going to, that is the, uh, the, the, not just the game changer, but that's the on switch for this. Otherwise, it, it's a nice par. A lot of AI is kind of smoke and mirrors and politics and ain't that, isn't that cute?

Isn't that cool? But if you can switch on your own data, that's when that's when it becomes truly, truly, truly valuable. Right?

Absolutely. Like game changing, valuable. Anyway.

That's right. Super. We, we try to keep these to 15 minutes we're way over.

I apologize. Um, but we've got even people in the waiting room for our next one here. Hey, man, I gotta have you come back on here.

We'll talk more about this. This, this is actually something we cover a lot here on Text Drunk tv. So consider yourself having an, an open invitation.

I would love to come back, Evan, it's always a pleasure talking to you listeners. Okay. Go check it out.

Thank you so much, Sir. Do Ayo io Io, make sure people go There, IO. Yeah, absolutely.

Alrightyy, Have a great day. Thank You, Suko. Be well.

Bye-bye. Bye Now.