Subscribe
Share
Share
Embed
upGrad Enterprise aims to build the world’s largest GenAI learning initiative to enable high-growth companies to embrace technology’s transformative business impact. Hosted by Srikanth Iyengar, CEO, upGrad Enterprise, the GenAIrous Podcast, will curate an exciting roster of global experts and guests, who are at the cutting-edge of Generative AI, and its varied applications in the world of business.
Welcome to the GenAIrous Podcast where we unravel the fascinating world of generative AI and its transformative impact on business globally. I'm your host, Srikanth Iyengar, CEO of upGrad Enterprise. At upGrad Enterprise, we're building the world's largest Gen AI learning initiative, empowering high growth companies to leverage cutting edge technology. Each week, join me and the roster of global experts as we explore innovations shaping the world of work as we know it. Let's get GenAIrous.
Srikanth Iyengar, CEO upGrad Enterprise:Hello, everyone, and welcome to another episode of the GenAIrous Podcast. We're taking you all the way over to the Bay Area today. You know, I have with me Sanjeev Gupta, a founder of a super exciting startup focused on Gen AI in the Bay Area. Sanjeev happens to be a longtime friend as well. Sanjeev, welcome to the show.
Sanjeev Gupta, Co-founder Kahuna Labs:Thank you so much, Srikanth. Excited to be on your show.
Srikanth Iyengar, CEO upGrad Enterprise:Wonderful. Wonderful. So listen, you're in an exciting space. You're right in the thick of it. So tell me a bit more about your startup. Tell our listeners what you're up to.
Sanjeev Gupta, Co-founder Kahuna Labs:Absolutely. Thank you so much, Srikanth. So what we are doing is we are trying to bring generative AI to transform support engineering. Support engineering is typically a function in, technical product companies, and especially in the b two b space where the customer requests are generally complex in nature. They are generally in the context of the customer usage, their configuration, their setup of how they're using your product.
Sanjeev Gupta, Co-founder Kahuna Labs:And it is an industry that has not gone through a disruption in a long time. So last year, when Gen AI came to this world, we looked at it, and it and it was, it was mind boggling what it could do. We sliced and diced into what are the capabilities that are now new toolsets in the hands of this world. Four things stood out. First was the capability of LLMs to do reasoning. The second is their capability to interpret language. The third is their ability to understand the context. And then the fourth is the body of knowledge that they can recite from. So with this new toolset, we started to reimagine what industries can be transformed. We looked at a number of different studies.
Sanjeev Gupta, Co-founder Kahuna Labs:McKinsey came up with the study, and customer support, customer service was at the top of it. Now traditionally, customer support is seen, more in the b two c world where the ticket volumes are high. The requests are relatively repetitive. And we also found that there's not a lot of gold data available, to train AI models on. We dug deeper, and we found that in the b2b world, it is a much more complex problem.
Sanjeev Gupta, Co-founder Kahuna Labs:The size of the market was mind boggling. The you know, on an average, every year, about $1,500,000,000,000 of, hardware and software infrastructure and applications are sold in this world. 20 to 25% of that is being paid every year, for support. And you'd be surprised about 20% sorry. About 10% of the support of the software engineers in this world, they are working on providing support.
Sanjeev Gupta, Co-founder Kahuna Labs:It's a huge market. And when the product vendors, when they are providing support, they, it's a you know, there's no rhyme rhyme and rhythm to it. A support engineer, they get a request. They, use their knowledge. They use their understanding of the product. They start asking questions. Many times, they refer to some documentation. Many times, they refer to serial engineers, but there is no one single platform that tells them which diagnostics to collect, how to troubleshoot a problem, and how to get to the resolutions. Companies have invested 1,000,000,000 of dollars in observability, which is a very, inside out view of the world. But when you start to take an outside in view of the world, which is what problems are my customers facing, how does it translate into the capabilities in my product, the configurabilities in my product, nobody has invested in that.
Sanjeev Gupta, Co-founder Kahuna Labs:So when we looked at generative AI, when we looked at this huge market, which could be disrupted with using generative AI, we felt we could do something transformational. And that is what formed the genesis of, Kahuna Labs.
Srikanth Iyengar, CEO upGrad Enterprise:Sanjeev, that's a fantastic story. I mean, clearly, you know, your background, your deep experience with world leaders in technology in the Bay Area, I think it's all led to this. So look, I think going back, you said a few very interesting things. You talked about, you know, reasoning, immediate replies, context, and the body of knowledge. So I can understand the sort of the problem statement.
Srikanth Iyengar, CEO upGrad Enterprise:But what really stood out to me was the size of the market. 1 in 10 software engineers is focused on troubleshooting. And, you know, as we all know, troubleshooting or enterprise tickets are not a sexy or cool part of the space. Most people wanna work on cooler stuff, but it is a big problem. And, you know, you didn't even touch on the fact that as we all know, there's a lot of backlog on tickets as well.
Srikanth Iyengar, CEO upGrad Enterprise:Because when crises happen, you know, the the backlog just keeps building up, and over the years, it builds up. So clearly, a huge untouched problem to be solved. But it's also fair to say that there have been different approaches to crunching this over the years even before Gen AI. So, you know, give us a sense of how that space has evolved and how Kahuna Labs is doing it differently.
Sanjeev Gupta, Co-founder Kahuna Labs:Look. There have been a lot of, attempts at solving this problem. And, it's very, conversation intensive world, when it comes to, you know, support engineering. And to extract knowledge from that conversation, typically, the technologies we had were either regular expressions or natural language processing, which were very fragile in nature. So as a format of the conversation changes, as the tone of the conversation changes, your ability to extract would go down.
Sanjeev Gupta, Co-founder Kahuna Labs:The accuracy would go down. And so at some point, the, the the vendors in the space, they started to give up on the ability to solve the tickets and then move to adjacent areas. Things like, could I predict the escalation, possibility of a ticket? Could I improve the, the ticket routing? Could I do a better matching between a support engineer and a ticket?
Sanjeev Gupta, Co-founder Kahuna Labs:But the fundamental problem still is that when a ticket is open, somebody needs to solve it. And the support engineers need help in navigating the tree of, possibilities to understand how exactly do I move move forward in this ticket. What diagnostics should I collect? And and for that, there is a ton of information available in in in how similar tickets have been solved in the past. You are right that, you know, this is not the most attractive career option for support engineers.
Sanjeev Gupta, Co-founder Kahuna Labs:The support engineers, their typical tenure in a company is less than a year. And, in a few years, they want to move to product development. They don't wanna stay in, you know, support engineering for their whole life. And so they need help. They're looking for help.
Sanjeev Gupta, Co-founder Kahuna Labs:They're not looking to be an expert in the current state of affairs for a given product in a given company. And so that's why we felt that this is an opportunity. Now what we have built at Kona platform is is a recommendation engine. And just like any, recommendation engine in the world, it has the complexities of retrieval, ranking, presentation, and so on. Imagine, you know, a problem coming at your hands, and you have examples of 10 different problems that have been solved similarly in the past similar problems in the past.
Sanjeev Gupta, Co-founder Kahuna Labs:The question in front of you is which of those steps should you follow? So there's a notion of ranking them. There's a notion of, providing optionality. There's a notion of providing order of, collection of diagnostics and so on. It's a complex problem.
Sanjeev Gupta, Co-founder Kahuna Labs:And so what we do is that we use generative AI as a tool, not as an end. And we use it primarily to extract the key value pairs from this unstructured conversations that happen between a support engineer and a customer. We combine that with the traditional methods of of clustering, of ranking, and and with that, we build something we call as the support knowledge graph, which is what we use to then resolve incoming tickets. And this approach has worked extremely well, at the customers that we have, you know, that we're working with. It is precise.
Sanjeev Gupta, Co-founder Kahuna Labs:The support engineers who, who leverage the troubleshooting steps we provide, they're able to respond to tickets in in in less than half the time. And slowly and slowly, we are getting to the point where we are able to give more and more precise troubleshooting steps for tickets who are which are, you know, 2 turns or 4 turns or or 5 turns, and and we're progressively moving, moving this forward.
Srikanth Iyengar, CEO upGrad Enterprise:No. Sounds sounds obviously clearly an evolution from how things was, you know, how things were done before. So just building on that from a client perspective, I can clearly see how you don't need probably the same size of armies or support engineers you needed before. You'll probably need less. But is it fair to say that given the fact that what you're building is a tool, it's something that'll make them far more productive and focus on more complex problems or even do something else. So the quality of the job that the lesser number of people do will be far better. Is that correct?
Sanjeev Gupta, Co-founder Kahuna Labs:Absolutely. Absolutely. Absolutely. And we're seeing that in our results. The the time it takes to resolve a ticket is coming down drastically.
Sanjeev Gupta, Co-founder Kahuna Labs:The time it takes for a support engineer to become effective is coming down drastically. The customers are becoming more and more happy with the level of support they're receiving. And so the capability of large language models to not only generate new con content, but also understand preexisting content is coming in very handy.
Srikanth Iyengar, CEO upGrad Enterprise:But at the same time, I know that, you know, initially, when you were discussing when you were in beta phase, you were discussing this with clients. Obviously, the business case was pretty apparent, you know, when you when you when you put it in front of a client. But still, you know, you and I talked a few months ago. Some companies has had hesitations. I know you made progress since then. We'll come to, that. But the initial hesitations, why do you think enterprises hesitated a bit on this particular? Is it data privacy concerns? What are the other concerns companies may have had that you've solved for?
Sanjeev Gupta, Co-founder Kahuna Labs:No. That's a good question. Look. When generative AI came into this world, every company at the board level said we need to do something in generative AI. Be that for improving our product, reinventing how we operate, or cutting down our expenses, improving our operational, operational efficiencies.
Sanjeev Gupta, Co-founder Kahuna Labs:Right? And they started looking at, at at vendors out in the world and there are not too many vendors. The because it's brand new technology and so they spun up internal teams. Internal teams have had mixed success and, and they continue to look for good vendors out there. Now in terms of vendors coming into the space, many of them have shiny demos.
Sanjeev Gupta, Co-founder Kahuna Labs:But those demos, they break apart as soon as they are faced with real world scenarios. So there is a level of skepticism in the enterprises when they talk to vendors. The second most important the second important piece is that, there is a level of skepticism about sending your, crown jewels over the internet to a large language model. It's a one way street. Once you have given your data to a large language model, the model has ingested your content. There's no way to take it out. And so, more than the business teams, the infosec and the legal teams, they are really, really wary of their data going out. So that's why we took the approach of deploying our platform, in the customer network, so that no, data goes out. And that, stuck very well, with our customers.
Srikanth Iyengar, CEO upGrad Enterprise:So you're sitting behind the firewall in that sense. So it's secure.
Sanjeev Gupta, Co-founder Kahuna Labs:We're sitting absolutely. Absolutely. We're sitting behind the firewall, and, and and we don't take any customer data out of their network. So they feel assured that their data, their the LLM, that we are using, all of that stays in their firewall. It is not gonna go out.
Sanjeev Gupta, Co-founder Kahuna Labs:But I think the most important piece really is the ability to show results. At the end of the day, they are looking for, for results. And, if we can demonstrate, significant, improvement over how things are done today, we win the deal.
Srikanth Iyengar, CEO upGrad Enterprise:Yep. So, just, but, let me ask a contrarian question. So obviously, it gives the customer comfort that it's a private LLM. It sits behind the firewall in that sense. But does that how do I put it? Reduce the effectiveness of the model because you're not learning from best in class across industries. You think it's a trade off? How do you handle for that?
Sanjeev Gupta, Co-founder Kahuna Labs:That's a good question. Look. We build telemetry. We extract telemetry out of our deployments of our platform, at each customer premises. And the telemetry really tells us about, what sort of extractions are being more effective.
Sanjeev Gupta, Co-founder Kahuna Labs:It tells us about the the extractions which are more accurate, than the others. So we extract only the the the level of knowledge that is helpful for us to leverage across customers. But when it comes to resolving tickets, tickets for 1 customer are based on their own product. They're based on their own knowledge base. They're based on their own past, you know, past tickets, and that doesn't translate very well from one customer to the other. So it's a it's a good mix of what is proprietary to a customer versus what we can leverage across other customers. Yep.
Srikanth Iyengar, CEO upGrad Enterprise:Yep. So looking, you know, over the last few months, I know it's been a very exciting time for you guys. So tell us more about that. How's that journey been and what's driven that adoption?
Sanjeev Gupta, Co-founder Kahuna Labs:At the end of the day, you know, support engineers, they, they don't stay very long at a company. They typically move companies around. So if you're successful at one company, the word spreads. The support engineers go to the next company, and they tell, yours they tell their story about how, you know, Gen a based platform, like Kahuna, really helped them. And so we are actually seeing inbound interest at this point.
Sanjeev Gupta, Co-founder Kahuna Labs:And so, so so the attraction has has has been has been very fast. It is something that I've never seen in my career before, for a small company, which is still building the technology, still getting our, feet wet. The level of interest we have seen from our customers is unprecedented.
Srikanth Iyengar, CEO upGrad Enterprise:No. Fantastic. And, look, I know firsthand that you hardly spend any money on marketing. So on the back of word-of-mouth, seeing this kind of traction is phenomenal. So but but tell me, how does the next year or 2 look?
Srikanth Iyengar, CEO upGrad Enterprise:Because clearly, the word-of-mouth will spread. The you know, if the value of your product is that clear, it will continue. So how do you see Kahuna Labs growing over the next 12, 24 months?
Sanjeev Gupta, Co-founder Kahuna Labs:Yeah. Look. That's a that's a great question. I think at this point, for example, our focus is more on on delivering to the customers that we have already signed up rather than signing more customers. So over the next, 12 to 24 months, first is we want to build the breadth of the platform.
Sanjeev Gupta, Co-founder Kahuna Labs:Ultimately, we are not in the business of just solving tickets. We are building a Customer Knowledge Graph. So a lot of companies talk about the Enterprise Knowledge Graph, which is the body of knowledge which is existing in the enterprise today as a whole. What nobody's talking about, the body of knowledge that is today spread across different tools in the enterprise, which really encapsulates a customer information. How is the customer using that product?
Sanjeev Gupta, Co-founder Kahuna Labs:What is their level of adoption at a user level? What kind of problems are they facing with the, with the product? Are they likely to renew or not? What other features can they use in your product? What can you upsell to the customer?
Sanjeev Gupta, Co-founder Kahuna Labs:And so, you know, our our overall goal is to build that overall customer knowledge graph for an enterprise, so that it can be used for not just, support, but also for, product adoption, for product education, and for overall customer success.
Srikanth Iyengar, CEO upGrad Enterprise:Okay. Yeah. Makes makes makes a lot of sense and, you know, wishing you all the very best. But tell me one thing. Obviously, the fact that, you know, enterprise productivity through optimization of ticketing is a problem that most companies have known about. But why is it that some of the big companies have not succeeded in sort of leveraging Gen AI to solve for this? Is there a reason? Because one would think that given their access, given their clients, this should be a no brainer for them.
Sanjeev Gupta, Co-founder Kahuna Labs:Yeah. That's that's a good question. Look. They've all tried. And I think at this point, if I look at the technology landscape, building a RAG based search is almost table stakes.
Sanjeev Gupta, Co-founder Kahuna Labs:It is almost given that if you have some documentation available in the company, it is going to be exposed through the right means, whether that is to support engineers or to the end customers directly, whether that is through a chatbot or a search interface or through ticket responses, automated responses, etcetera, that would be done. That that's table stakes today. But when it comes to going deeper than that, which is the the the rest of the 90% tickets for which there is no documentation available, where the support engineers are relying on tribal knowledge. They're relying on the experience of senior engineers. It's a hard problem to solve.
Sanjeev Gupta, Co-founder Kahuna Labs:It's a hard problem to solve because for a given company, investing in building a platform that is deep enough to do these knowledge extractions, build accuracy, build a whole recommendation system is a ton of investment. And so when you put together, you know, 50 different companies, which have the same problem to solve, you have a a definition of a platform in front of you. If we build that, customers will come, companies will buy that, they will use it, they will leverage it. So it's it's a tough problem. It requires deeper work.
Sanjeev Gupta, Co-founder Kahuna Labs:It it has only now been possible to solve that problem, with LLMs given their ability, to do those 4 things that we're talking about.
Srikanth Iyengar, CEO upGrad Enterprise:And and look, I think that's you're right. You know, all the experiments that are happening in the space or the innovations require investment because you're consuming a lot of cloud compute. You're consuming, you know, tokens left, right, and center. All of these costs add up. And I know that, you know, if you go back a year, you know, while VCs weren't really investing across the board, Gen AI was an area where a lot of investment started. Some of it was FOMO as well. But while the investment continues over the last 3 to 6 months, there has been a tempering of the animal spirit, if I call it that. There's a little more caution. There's a little more diligence. How do you see that playing out?
Sanjeev Gupta, Co-founder Kahuna Labs:With the Gen AI every industry has to transform it's a megatrend it is not a mini trend right over the next few years almost a decade we will see almost every industry go through transformation. There will be small ebbs and flows. There will be some downs and ups, and there will be, euphoria, and there will be rationalization along the way. And so we are going through a very similar cycle. VCs, startups, large companies, everybody went through a euphoria last year. They heard about generative AI.
Sanjeev Gupta, Co-founder Kahuna Labs:They dreamt of companies. They dreamt of problems that can be solved. Some were, very rational. Some were almost in the in in the land of magic. And as people started, started using it in real life scenarios, they're now figuring out what is real and what is not real.
Sanjeev Gupta, Co-founder Kahuna Labs:And so for the most part, a lot of companies, a lot of investors, they are taking a wait and watch approach. They want to invest in the winner, and the winner is not clear yet. You might find, 10 companies in the same space where whose tagline is exactly the same. They say the same thing. 3 of them are trying to create some magic. 2 of them are trying to actually solve a problem. 2 of them actually have built the technology, and 2 of them are actually customers in front of them. And so they are waiting and watching and seeing who the winner is. In every industry, when VCs invest, they want to invest only in the winner. They don't wanna invest in, in the number 2 or the number 3.
Sanjeev Gupta, Co-founder Kahuna Labs:And so, I think the skepticism is not about what Gen AI can do. The skepticism of is about who is going to be the winner. And so that skepticism is gonna, is is is gonna change. But, also, I think it is a function of how the public markets are are performing. Over the last 1 year, public markets have performed very well.
Sanjeev Gupta, Co-founder Kahuna Labs:And so, the money that, the the drive out that investors have been sitting on, they are putting a drive out in into public markets. Yep. And they're getting really good returns. And so that that is also gonna change. So I think overall for generative AI, there's this has been a little bit of tempering. It is coming back up, but the megatrend is really that generative AI is going to transform industries. And any any product company who is not fundamentally rethinking how their industry is gonna change, with generative AI, is is gonna find a hard time in the next couple of years.
Srikanth Iyengar, CEO upGrad Enterprise:Absolutely. And look, I can speak for us, for Upgrad. You know, we are leveraging Gen AI all over the place in terms of the kind of content we generate, how do we handle customer experience, how do we reduce the cycle time to getting, content out to customers, etcetera, etcetera. So, completely I mean, if we weren't doing that, we'd be missing a trick.
Srikanth Iyengar, CEO upGrad Enterprise:We wouldn't stay ahead of the curve, which which we pride ourselves on. So I can completely relate to that. But let me let me shift gears. I think, you know, the exciting space But coming back to your direct consumer, and I and I don't mean the enterprise customer, but you deal a lot with, support engineers. And there is a lot of debate about human versus AI or human and AI, whichever way you look at it. And what you're, sort of building is a tool that's, in many ways, replacing the support engineer in their current shape or form. So how do you see that human AI equation playing out in the space that you focused on?
Sanjeev Gupta, Co-founder Kahuna Labs:Look. If anybody thinks that suddenly AI can replace humans, that is not possible, with today's technology. Things have to evolve. Things have to build towards that. So today's focus is to assist the humans.
Sanjeev Gupta, Co-founder Kahuna Labs:It is to provide them the tools, provide them the intelligence, from AI, to do their job better, to be more effective at what they're doing today. And the example I take is for self driving cars. You know, so over the years, slowly and slowly, car companies have been building more and more capabilities to get closer to self driving. You know, slowly and slowly, they're building more and more logic into into the software rather than into the hardware. And the hardware, component is being more focused into acting more as sensors.
Sanjeev Gupta, Co-founder Kahuna Labs:And the sensors bring in information. More and more data is being collected. The software is being tuned to continuously improve to get us to the point where you can get to self driving cars. The same paradigm applies in any AI driven system. You start with, figuring out where are the humans spending time, what are the different steps that they perform on a day to day basis, where can you provide intelligence in those steps, where can you bring in automation, and you slowly and slowly start to chip away at those steps one at a time.
Sanjeev Gupta, Co-founder Kahuna Labs:At some point, the, two things happen. One is you improve the accuracy of your steps that you're performing using AI. Second is that you continuously expand the boundary of of the steps that you're doing. You you you start to automate more and more steps. And that's exactly the approach we have taken.
Sanjeev Gupta, Co-founder Kahuna Labs:It's a journey. It is not a one day journey. It's not a 1 month journey. It's not a 1 year journey. It's a long term journey to get to the place where, the AI is in the driver's seat, and human intervention is more an exception. We are slowly getting there, but it's a long journey. And I and I and I and I think it is probably going to be a very similar trajectory in every industry that starts Leveraging AI.
Srikanth Iyengar, CEO upGrad Enterprise:Absolutely. And, you know, I must share something on a lighter note with you, Sanjeev. This weekend, as you know, the long standing CFO of one of the tech leaders stepped down. He's taken a break. Yes. And so there was a slightly provocative article in the Financial Times about the fact that the next CFO of 1 of these companies could be an AI bot. You don't really need a physical CFO. And, you know, it was an interesting, sort of editorial. But the comments were even more interesting because one of the comments said that, you know, if you have a bad quarter, you can sack the CFO, but not an AI bot. So that was quite interesting.
Srikanth Iyengar, CEO upGrad Enterprise:So so there are some things I think humans can do that AI just cannot. So yes. But but, no, Sanjeev, listen. It's been a fascinating conversation. You know, clearly, you guys are at the cusp of some amazing work, and, Kahuna Labs seems to be on a phenomenal growth path. So, you know, we wish you all the very best. With that, thank you so much. And, thank you for listening.
Sanjeev Gupta, Co-founder Kahuna Labs:Absolutely. Thank you so much, Srikanth. Thank you for having me. Enjoyed talking to you and look forward to staying in touch.
Srikanth Iyengar, CEO upGrad Enterprise:And that concludes another episode of the GenAIrous Podcast. We are very grateful to our guests for their time and expertise. A big thank you to our producer, Shantha Shankar in Delhi, and our audio engineer, Nithin Shams in Berlin for making magic happen behind the scenes. Join us next time, and don't forget to subscribe to GenAIrous wherever you listen to your podcasts.