[00:00:02] Rajan:  AI is reshaping businesses, and ignoring it could leave your industry disrupted. I'm Rajan, and this is Pivotal Clarity. We talk to those building or using AI, founders and engineers with real world experience. Our aim is just to cut through the hype and see where AI is truly making an impact. If you're a business or following tech trends, these conversations offer clearer insight than most of the press. Let's get into today's episode. Welcome to Pivotal Clarity and AI podcast. In this episode, I have Prasanna, my partner in crime, cofounder at Upeka. So far, we've been bringing in guests. We're working on different areas, whether it is robotics or on AGI and other places. I thought it'd be good to sort of bring in Prasanna and then have a conversation around what is happening from an enterprise perspective in AI as well as what is happening in enterprise for Indian founders. So, Prasanna, welcome to the show.

[00:00:59] Prasanna:  Great to be here, Rajan.

[00:01:00] Rajan:  Alright. So let's jump in. Prasanna, what do you think has happened since the chat g p d moment? Uh, what does it mean from an enterprise perspective? And it's been close to about, uh, 2 years now in November, the the chat GPD moment. What has happened for a b to b founder to actually make sense of?

[00:01:17] Prasanna:  Yeah. I think, you know, it's so hard to even think about how far we've come. It's phenomenal. I think goal posts have kept shifting in terms of what can happen with this and what this technology brings to the table. And I think for a lot of technologies, enterprises have been very slow in adopting or trying to adopt. But this time, I'm hearing of very large enterprises having use cases which were previously very manual and very time consuming, which are actually not being attempted, but have now been automated in production at scale and saving them, you know, 1,000,000 of dollars a month. So I think, you know, just looking at the last 2 years and seeing what the new Gen AI stuff has brought to the table, and especially the UX of chat as something that has enabled everybody who is reasonably computer savvy to start using it. I think the combination of that 2 has really changed a lot of things for everybody.

[00:02:13] Rajan:  What do you think is has been the biggest surprise for you in the last 2 years?

[00:02:16] Prasanna:  I think the last couple of years, everything has been surprising. I think started with, you know, chat and just text as a format and seeing how that would be very powerful. And then seeing images and everything else around that and generation has been phenomenal. And more recently, literally being able to power an entire three d world straight from the AI instead of having to build the world. It's very, very raw. It's It's basically, you know, nowhere close to anything real. But all of these have been mind blowing. And I think the multimodal AI, which uses images, which uses voice, which uses text, which is able to pull in all these things to make sense, I think that is phenomenal. When I saw 1st multimodal demonstration where they took a photo of a cycle and a cycle seat and said, hey, how can we adjust this? And it gave instructions back after looking at that picture, identifying what kind of a bolt it was and how would you loosen it, and then how would you tighten it. That I think is like was really mind blowing to me because that kind of support in your real world with looking at the real world, hearing your question, and being able to answer that, that was to me one of the mind blowing moments of, like, wow, what else can this do?

[00:03:31] Rajan:  So basically, those things that were not possible or, like, you couldn't imagine those things before. Now when you see those things happening, those are the mind blowing moments. Any specific from an enterprise perspective? Maybe we should chat about what are some mind blowing things that we've seen founders do in enterprise context.

[00:03:49] Prasanna:  I think one of the use cases, and this is not a founder who told me, but an enterprise use case nevertheless, was that they have this particular enterprise, very large enterprise, has people coming in with credit cards and making payments. And when these people come in with credit cards and make payments, some of them later file charge backs. And so a charge back is when the person who's buying something says, hey. This thing didn't work or I want to done it, and they're not taking it or whatever, and they basically reverse the charge. And for each of these chartbacks, the enterprise needs to look at the chargeback, identify what was bought, identify what is the issue, did the customer raise the issue. They would write a letter and they have to send it with proof, if any, saying that, hey. No. We did our job. You should not honor this chart back. And if they don't do that for all the cases, then their rates will go up. That is the merchant discount rate, which the credit card company, that's their commission on the transaction. That will keep going up. Right? And earlier, they used to have teams of dozens, if not 100 people looking at each of the chart backs and trying to get the evidence, put it together, see if it makes sense, and then file the case. Because whether they're going to win the chart back or not, they have to do this stuff. Right? Now they've used Gen AI to automate this whole thing, and they can get an image from a camera. They can get the receipt. They can look at the item, and they've made Gen AI do all of this stuff, write the letter, send it back to the bank for the chart, the credit card company for the chat. Right? And they are saying that they're saving 1,000,000 of dollars a month doing this, and this is already in production. Right? And this was something that, you know, only people could do. It would have been very hard to automate this because each chart back has different information. Each thing that's bought has different information. You have to put that together, create some evidence logs, put that together, and make a case for why the chart back shouldn't be there. This is pretty, like, complex stuff that an analyst kind of a person would be doing. Right? And they've been able to automate the whole thing with AI.

[00:05:45] Rajan:  Fresno, what's your take on SaaS versus AI? The changes or the shift that is happening.

[00:05:51] Prasanna:  Yeah. That's a good one. That's a good one. I think SaaS became category of software, and what's happening is that SaaS has a revenue model, a recurring revenue model, is probably going to stick around for some things. But SaaS as a model without AI and without a lot of AI, that is going to become less and less powerful. Because many of the use cases that SAS solves today, they are little focused. They're a little narrow. And before the SAS comes into play and after the SAS comes into play, there are things that humans need to do in order to make whatever job to be done happen. And what's happening with AI is that many of these jobs to be done before and after are higher value. And those higher value things are now, some of them at least, are possible with AI. And so it is more profitable for me to automate those parts which are higher value rather than just the SaaS part of it, the old SaaS part of it, which isn't very narrow. The other thing that is happening is that many jobs to be done were broken down into silos, And those silos were broken down in a way where individuals could do each of those things. And then the job that an individual could do was then mapped into a SAS tool, and that SAS tool would do that part of the job. What I'm seeing with AI and what a lot of people are seeing with AI is that that kind of a silo breaking for AI is not productive. In fact, if you go to the higher level task and without breaking it down, you take that whole task and see which parts of it can be done by AI, which parts can be done by automation or software. That is a much better way of looking at it. And so one of the things that I think is happening and is definitely gonna happen more of is a lot of SaaS tools have been working on this silo, which was created in order to enable 1 person to be able to do that part of the job and to do it repeatedly like a factory line automation model. Instead, AI tools are being built that does the whole job or at least large portions of that job with inputs on the fly from 1 or 2 people who are more expert at understanding what is needed at the top. So they know what they want. So they tell what they want to the AI and tune the AI to get what is required end to end rather than the pieces. Some of the SaaS companies, the mistake that they're making is they're trying to add AI into this silo, and that is not working very well. Right? But the AI companies that are saying, hey. I'm gonna reimagine the whole piece and try to do it end to end. Those folks are doing really well. Now the caveat there is that maybe the AI today is not powerful enough or good enough to do that. And that's where the newer things that have been done, whether it's the chain of thought stuff or the reasoning stuff or whatever you call it or the agent stuff and things like that, that might provide the leg up, or more powerful models that are able to do this better will provide a leg up. Or in some cases, it is about the data accuracy, and therefore, RAG or something else comes in. But I have no doubt that very soon we'll start seeing this end to end stuff where basically an expert or an author is able to say, what do I want and how do I want it to be? It's telling the AI what to do rather than a lot of different people cooking small bits of these things in silos and then trying to put the whole thing together.

[00:09:03] Rajan:  Correct. That means you're not taking the bait on a SaaS dead. That is the, like, thing that is going around right now. Right? So is people are talking about a SaaS dead.

[00:09:12] Prasanna:  Absolutely. So I will take that bait in that sense. Right? I think SaaS is dead in the form that it used to be. I think siloed SaaS as a tool for an individual to do one specific thing. I don't think that is going to survive.

[00:09:24] Rajan:  So I also heard the following way to sort of describe how people are building it. So you mentioned about, are you doing it in silo? Are you doing it like, you know, adding it as a, like, you know, foundation? Are you putting it an add on? The thing that I heard is that people are saying that, you know, many founders or many companies are building AI into their offering the way you would add floor after a cake is as baked. You can't do that that way. You have to actually do it much before maybe you call it as gen AI native or AI native. But you have to think about it from grounds up. What's your take on that?

[00:09:54] Prasanna:  Absolutely. I think by and large, I completely agree with that. I think that, you know, from a technical perspective, one could argue that the previous generation shift, which is from desktop software or client server software to SaaS, which is hosted on let's just call it multi tenant hosted on cloud is not necessarily technically too difficult to imagine. But most of the companies that were client server or desktop didn't survive the transition because there was a business model shift as well. There was a GTM shift as well, and that changed how things happen. I think we're in the early days of the SaaS to AI software shift. And, technically, it doesn't sound that like adding AI to a SaaS product or a software product should be too hard or impossible to do. But I think that what AI allows is to look at pieces of work that humans did and do it in a better, more automated fashion. And that way of looking is something that is a mindset and hard to change for people. And so people who start looking at it from that perspective as, hey. What are humans doing here which can be automated rather than, you know, which parts of this work today can be automated, which are already in software or closed software. But look at the entire gamut here.

[00:11:07] Rajan:  Elimination before automation. What I'm hearing is say is that there are places where, you know, things should not even be automated. They are to be completely eliminated because you're thinking about things in a completely different way. So when people add it as an add on, then they only think about it as, okay, here are 5 steps, and how do I automate it? But when you think about from a fresh mindset, you will say that here are 4 steps that we should eliminate, and here is just 1 step that needs the automation and things like that.

[00:11:30] Prasanna:  Because people are already inside the box, and you're looking at the box, and you're looking inside the box and saying, hey. Can AI do something inside the box? But really today, what has happened is that stuff is already being done by software in some way, shape, or form. And there's stuff before that box, outside that box, or after that box, outside that box that AI might be able to actually deliver the highest impact.

[00:11:49] Rajan:  What do you think of some of the biggest misconception that you've seen whether generally or, like, you know, amongst big companies or start ups?

[00:11:57] Prasanna:  Uh, it's a good one. I think one is not reimagining the process and trying to just add on AI like you said. I think that's definitely a big one, retrofitting AI. Another is and I like this metaphor. I think it's become quite popular now. Is when you want to do things that are precise, mathematical, specific, step by step, and that need to be precise, then by all means, please use software and don't use AI. Trying to use AI to add the 2 plus 2 is probably not the right thing to do. Trying to use AI to figure out which two numbers need to be added in a particular problem. That's a good use of AI. You're going to struggle to use software to do that. But actually adding those two numbers probably, like, build a system that will take the two numbers that AI found and, you know, which operation needs to be done from for that problem. Put it into a calculator, solve it, get the answer back. I think the separation of what needs to be done by AI and what you need to build software for, I think that's where, again, the AI first companies are doing a really good job of having both. 1 is not enough. That is, I think, the number one misconception there. People think that AI will do everything, and that is completely an incorrect.

[00:13:10] Rajan:  So 2 weeks ago, I met one of our founder and he was saying, what is so different about AI? I mean, we've been using AI for the last 5 to 6 years, and we ran analytics and we were able to sort of run some ML algorithm from the data that we collected and then we showed insights. So what's the big difference between, like, what we were doing before versus doing now? And there's, like, a sense of frustration in his voice saying that there's, like, you know, AI that was done before. Is this really hype? What would you say to him?

[00:13:34] Prasanna:  Yeah. I'm very lucky. Right? I'm married to a PhD in machine learning, AI, whatever you call it, and she's been doing that since 2012 or 11, some somewhere in that time frame. Right? I've seen what machine learning algorithms are at a close distance. And so the machine learning algorithms were all by and large single purpose. So you design them for a purpose, and then you use them for the purpose. And so previous generations of analytics products, for example, would be designed to apply an algorithm which and that algorithm will get only a particular type of insight. So if you want to identify highs and lows or you want to identify trends, it will identify only a certain type of trend that it is programmed to identify those kind of trends. Now the new type of AI that is available is a more general purpose AI, and the kind of insights that it can come up with are not constrained by what you are trying to make it come up with. And that is a key difference. So when you start doing this across text, which is what it is mostly very good at, and for images, which is also what it's getting good at, the insight that it might come up with is something that may be quite unexpected. And so that part of it is where Gen AI is getting really good at. And so one is to say, hey. I'll run this machine learning algorithm on this particular data, and it can provide trends of these 3 types. I'll take that. But now earlier, what these folks would have been doing is they would have taken this. There would be an analyst who now writes a report with these trends that the machine learning algorithm has identified and write a conclusion, write some recommendations, and things like that. Right? What Gen AI has done is that analyst job of looking at the trends, identifying which trend might be the most salient for this period, writing a conclusion, writing recommendations, you can have AI, the current AI, whether it's from anthropic or open AI or anybody else, you can actually get it to do that, and it will do a pretty good job. And so doing that might actually increase the value of what you're doing substantially because let's face it, lot of these trend analysis, lot of these insights that were derived and stuff like that are were a little too crude for the end business user who's nontechnical to be able to take action on. Right? And that meant that lot of these trends, a lot of these analysis were just left on the floor. But this step of being able to convert it into business talk for that particular organization with their, you know, custom information embedded, that increases the value a lot and makes it, like, really actionable. Right? I can look at that and take some action. Now, of course, there is hallucination. Of course, there are a lot of problems there. But that stuff is just getting better and better and better.

[00:16:15] Rajan:  So, like, what was special purpose before is it become general purpose? And then the power of the solution that is coming from the general purpose AI is just so good that, you know, it is far better than the special purpose one that was deployed before. Therefore, we are calling this as a generational shift. It's almost like, you know, v one and v two end. Because that it feels like, you know, such such a jump, it is a different one. And that leads to possible.

[00:16:39] Prasanna:  Yeah. I do want to say that the ML algorithms are also critical. What the ML algorithms does, the open AI API or the anthropic API cannot do, will not do well. Right? And if you have a large amount of data, it literally will not be able to do it because it's not designed for that task. Right? And this again is where people need to get clarity, and this is a great example of what we were speaking about before. When you want to do something mathematically, you want to take, you know, a 1000000 data points and look at what trend is there and stuff like that, please don't do that with open AI or anthrope APIs. Right? Please write custom regressions, SVMs, whatever you need in order to take that massive amount of data and get things out of it. But once you get things out of it, making that far more human friendly is something that your algorithms will not be able to do, but the new types of AI, the Gen AI stuff will be able to do really, really well.

[00:17:33] Rajan:  Yeah. It makes a lot of sense. So one of the things that has been discussed the last couple of years, 2 years, or even more is is that this feels like a shift. Right? The platform shift. And then the platform shift always is a question of who does this platform shift enable. Right? Does it enable the incumbent versus enable the startup? When this shift was initially happening, I had a view that, you know, maybe the incumbent has advantage. But then if you see, like, now this year alone, maybe about close to $50,000,000,000 of investment has gone into startups. I'm talking about the global volume and not the India one. India one is perhaps only $250,000,000 in the last 8, 10 months. What's your view now seeing now incumbents are innovating, what startups are doing, who has the advantage, who has the edge? And maybe if I were to extend that question a little further, how to win? Like, how does incumbent win and how does the startup win?

[00:18:23] Prasanna:  I think incumbents have a lot of smart people working for them who they pay a lot to figure out how to win. So I don't need to give incumbents any advice. They're not paying me yet. Right? But on the other hand, for startups to win, I think, you know, we need we all need to think that through. Right? And to me, one of the beautiful things about this platform shift and this generational shift that is happening is that it's just opening up a ton more of opportunity. Right? It's opening up such a huge opportunity that there are gonna be large, massive opportunities for everybody. I don't think we should operate this from a zero sum mindset of, okay, if incumbents win, then start ups will not win or vice versa. I think, like, there is playing room for everybody. There is gonna be massive, massive amounts of money, value being created, true value being created that was not there before, which everybody is going to get, you know, pieces of that pie if they make sure they're doing it right. As an example, if I have to take something very fundamental, there is not enough health care available at a point geographically that most of the world population can access. It's literally not there. Can we train enough doctors? It's not clear whether we'll be able to train enough doctors in the time frame that we need. Right? So it's not clear whether we'll be able to train enough doctors in the time frame that we need them in. But what's happening with AI is AI is rapidly getting better at making diagnosis based on data where somebody who is a nurse or a trained health care professional, but not necessarily trained to the level of being a doctor, will be able to use AI and make a better diagnosis than people would have gotten otherwise. Right? And this is going to be for about, I don't know, anywhere between 3 to 5000000000 people on the planet. And so that is the kind of value that's being created. 3 to 5000000000 people who have not had great access to health care before will now get better access to health care because of AI. And this is gonna happen very fast in the next 3 to 5 to 7 years. It will transition so fast. We won't even know what's going on. Will this all happen because of incumbents? No. Because there will be startups which will build mobile apps, which will help a nurse somewhere remotely to just use that app and make a diagnosis because there's no doctor around for, like, a few 100 kilometers. Right? And which incumbent is gonna take that risk and do it? Probably nobody. But is there a startup that's gonna take that risk and do it? Yes. So that startup might win in those kind of cases. And then over time, that use case might become allowed. Right? Because there's no going back once this happens. So there are gonna be new use cases which create 1,000,000,000,000 of dollars of opportunities. There are gonna be existing use cases which are going to create 1,000,000,000,000 of dollars of opportunities. In my mind, where there's a lot of reimagining that needs to happen, start ups have an advantage. Where there is a lot of relationship or top down kind of a sales in an enterprise which is required, then existing enterprises, especially the tech hyperscalers, AWS, Microsoft, have a massive advantage in being able to sell stuff top down, Oracle, SAP. Right? So they will acquire startups which are able to reimagine stuff, and they will then reimagine stuff at scale. That's my take on it. Consulting companies will also be able to get in on doing this at a high level. But I think a lot of consulting companies may struggle to make the shift to being tech led and AI led rather than being consultant led because their DNA is, hey. We'll throw people at the problem in some way, shape, or form.

[00:21:35] Rajan:  An underlying pattern that I heard you say is is that, you know, start ups will win in places where incumbents can't compete or won't compete. That's that's a place where you go and pick those niches or, like, crevices or gaps, if I were to call them, and then go and play in those spaces where incumbents are like, look, I don't wanna take a risk or I don't wanna actually play in this space. And that itself can become like a large opportunity is what you're saying. And then wherever incumbents are playing, what you're saying is that there, if you go and reimagine the experiences, that's something that, like, you know, incumbent may not be the best person to do that. But however, wherever it involves, like, lot of people to get aligned to make a decision to do rollouts, wherever relationships are key and are option to happen. Where existing relationships are there with incumbents, that's where, like, you know, incumbents will have an advantage to kind of win. How does this change GTM, Prasanna? Like, what they are ready to take that on?

[00:22:25] Prasanna:  If they're ready to also take that on. If they're ready to take it on as well. Because flip side of having that relationship is that you may not want to go and tell them, hey. You need to reimagine. You may want to preserve status quo. So there is a flip side to that relationship as well.

[00:22:39] Rajan:  Correct. So as part of this, how does it change GTM? See, one of the big changes that had happened, uh, was, like, you know, selling to assisted buying, which led to SaaS, like, 20 years ago, which was pioneered by Salesforce, and then, you know, back in India with Zoho and Freshworks. And the buying behavior change when, like, you know, change that led to change in go to market. And because of all these changes that are happening, what are the go to market changes that you're seeing, and what are the emergent ones that you're seeing?

[00:23:06] Prasanna:  Yeah. That's a great question. I think that is the biggest question for entrepreneurs going forward. And I think definitely there are a few different categories there with different strategies that are going to work, but in each of these categories. And I think in the prosumer space, that's the one which is the most noisy and where, you know, a lot of things are popping up. And I think there, you know, try before you buy is becoming something that's very, very, like, critical. You have to have a working demo. You have to be able to showcase how well the AI works. It has to be mind blowing. It has to be something that people go, yeah, magical. And people go like, holy crap. I wish I could do this right now, and I I can't even believe that I'm doing this right now. But, of course, in a few weeks, it becomes, oh, yeah. Of course. Yeah. This is, like normal. Right? So that kind of thing is what it seems to be moving units. Right? It's that's what seems to be selling stuff. But on the other hand, in the enterprise level, I think that is still very much out from a jury perspective in terms of what will actually work. Because this kind of reimagination, this kind of process shift, this kind of large scale end to end kind of change is not something that is easy to convey. It's not something that's easy to even understand what needs to be done. It's not easy to understand where the value is. And so a lot of, I think, selling is going to happen initially at least by domain experts who are able to both convey the credibility that, hey. I know this process end to end, and I can tell you that there is a way to reimagine this to be way, way better. And I am doing it, and therefore, you know, you can trust me that I will take you from point a to point z in a way that is, like, far, far better for you than what you were doing yesterday. So I think domain expertise is going to be very, very critical in selling that. And I think once that sales process is figured out, what are you selling, how are you selling, all of that stuff. Then in my opinion, the challenger sale kind of methodology is what will be required by and large to say, hey. You know, what you're doing doesn't work, and I'm gonna challenge your status quo. And I'm going to tell you there is a better way to do it, and I'm gonna prove to you that there is a better way to do it, and kind of take that on and do it. Right? The more the sales processes and sales models that are little more status quo, little more, you know, just incremental shifts, little more, you know, fit into your existing ecosystem kind of things, I don't think those sales processes and sales models will work as well for this kind of transformational shift.

[00:25:38] Rajan:  What are you telling SaaS founders that are coming to you and asking for, hey, what do I do with this AI shift? What has been your top 2 or 3 advice in the last 1 or 2 quarters?

[00:25:48] Prasanna:  Yeah. The first thing I tell them is, man, nobody knows what is going on. Right? I think people seem to assume that somebody knows what's going on, and then they'll let me figure out who knows what's going on, and then let me do what they are doing. That worked in maybe SaaS and software and things like that where the pace of change was slow. So stop looking for somebody who knows better than you. That is, I think, the number 1 lesson that a lot of founders need to learn. The number 2 lesson is therefore, the only way to figure out what works in this situation is to actually try something, take it to customers, see if it is them, see if there is value created for them, understand, be able to do that cycle quickly, do that cycle more and more, and be able to take that cycle and what comes out of it to the next level. And, you know, you have to do it yourself. You cannot look for a superhero above you or, you know, a leader above you who is gonna tell you which way is gonna evolve because nobody knows how this stuff is evolving, how fast this stuff is evolving. What is going to work next? What is not going to work next? Nobody has a clue. Right? Everybody is making stuff up. Everybody is, you know, rolling their own path, and everybody is maybe, like, half a step ahead of you, 1 step ahead of you, maybe 2 steps ahead of you. But nobody is there yet. And so you cannot look for what is a dare and say, yeah. Yeah. I will go there. No. Because that there itself is not very stable. A company that might be at 20, 30,000,000 today in revenue from AI products may very well be completely not required year from now or 2 years from now because it's completely baked into an existing incumbents product or the AI platform itself or something shifts so dramatically that, you know, that value creation is not the main important thing anymore. Right? So I think founders, in many cases, are looking for certainty and never existed. But today, of all times, it doesn't exist even more than it never existed.

[00:27:47] Rajan:  So 2 days ago, I saw Jake Sapper, a friend from Emergence Capital, post this on LinkedIn saying, shouldn't we call the services a software as AI services because it is more around outcome driven? Because I went and commented. What is your take? I know it's a very seemingly trivial sort of a question, but, like, a lot of founders come to me and then say, hey. This service is a software. Is this there? Can you give me example? And to your previous point, you know, can you give me playbook for how to make services as a software work? Our take your take on that.

[00:28:18] Prasanna:  Absolutely. I think AI in the way it is today in the world map kind of a context, it's somewhere between Genesis and custom. And for those of you know and understand world maps, when something is in Genesis and custom, it requires a lot of hand holding in order to make it work. And it's evolving rapidly, and therefore, keeping up with the evolution also requires a lot of work. And, of course, the common frame that data is very hard to keep leashed, if you will. Right? Because new data keeps coming, that new data may not look like the old data. You need to still keep molding it and molding it to get it to where you need to be so that you can then use AI on top. For those of you who don't know what modeling mapping is, please learn it. Uh, in the age of AI, modeling maps are gonna become more important than ever before. Your AI can help you build modeling maps as well. But coming back to your question, because AI is in Genesis and custom, a lot of services are going to be required in order to make AI work for you apart from cleaning the data for you. Make AI work for you and keep AI working for you and keep improving the outcomes that you want as AI keeps improving. So because of these reasons, services are very, very critical part of delivering AI. That is one part. The second part is GTM reason in my opinion, which is that people have line items to buy services. And these line items to buy services are typically on outcomes. So you will hire an agency. You will hire a contractor. You will hire resources as in a services model where you're paying month on month or you're paying quarter on quarter or whatever it is for a certain number of outcomes that are delivered. And so that mental model is already there. So if you build AI products and try to go sell it as a product, as a line item, then the way they buy that product is very, very different, and that process is very tricky, hard, involved, and so on and so forth. And on the other hand, the AI product may not be delivering the entire value that the customer needs, and things are shifting rapidly. Right? So what most folks are trying to do is there is a layer of people who are delivering a service to the customer, and the customer understands what they are getting because they have bought services before, and so they're paying for those outcomes. And they say, sure. For this outcome, we can pay this much. And they're used to getting it delivered by people, so you kind of make it look like it's being delivered by people. You have some people because the data is gonna be funky, the outcomes are going to change or need to change, and the AI itself is changing. But then you build your AI under that as a product, which is evolving, doing stuff, doing more, changing over time, keeping the data clean, all of that stuff, and you deliver that now through AI rather than through just people. And this combination, especially from a GTM perspective, I believe, is gonna be more palatable for most enterprises than trying to go sell them a product. Because the reality is that over the last many years, couple of decades, lot of people have been telling these people, buy my product, your life will change. And they bought all these products and their life hasn't changed. Versus they know that when they buy a service, they get something done end to end. And at least that part of it, they don't have to worry about so much. Right? So that I think is going to be a big shift. And I like the fact that, you know, maybe we should go and call it AI services or agent services or whatever, and maybe a better name than services as software because services software is like a mouthful, and I find it hard to say many times.

[00:31:40] Rajan:  So the part of the question, Prasanna, was also on the name. Right? So how do you think about, like, you know, which name will evolve? Then to your point about, like, wordly map, which is Genesis to custom and then custom to product. And the custom to product happens only when the category name gets defined. Right? And because the budget for headcount and the budget for tools are merging. Or, like, you know or in the custom case when you don't think about it as a category name, but you're really focused on the outcome and then really custom build. The GTMs are much faster. So based on these 2 aspects, you know, where do you think and how do you think, you know, the name will shape up? Because I'm sure, like, you know, many people are battling for calling that name. Somebody calling it as outcome as a service. Somebody has software and services. Services as software. AI services. And what's your take on, like, which names will shape up better and why?

[00:32:25] Prasanna:  I like AI services. I don't like outcome as a service. I don't like software a service as a software simply because the acronym will become too close to SaaS, and I think it'll create a lot of confusion. I think agentic AI services or, you know, AI agent services are also possible, but maybe too close to it being as as well. Right? So it has to be a name that works. It has to be an acronym that works better. So let's see.

[00:32:51] Rajan:  Yeah. Also and those who actually can frame the problem part of this better. Right? One aspect of this is it.

[00:32:56] Prasanna:  I think we are still 2 to 5 years away from that, in my opinion, for it for it to become a category.

[00:33:01] Rajan:  Yeah. And the whole SaaS was something that took off. The previous version of that was ASP. So, Prasanna, how were your own personal workflow changed? What are tools that you use? What are AI things that you use in your daily life?

[00:33:12] Prasanna:  I think I'm, like, super boring on that front. I use cloud. I use gbt on my phone. Very typical use cases. Kids want something, take a photo, ask it to describe what's happening in the photo. My, uh, daughter had a bee sting. So took a photo of the bee sting and asked, okay. What what do we need to do to get the sting out? Stuff like that. Right? So very boring use cases, I guess, but lot of usage of cloud and OpenAI GPT per both work and and hope. I think the most recent thing was my daughter and son had to study something. So I had GPT make an acrostic, small alliterative acrostic for them to memorize more easily.

[00:33:48] Rajan:  Any final takeaways for founders that you wanna leave with?

[00:33:52] Prasanna:  Uh, absolutely. I think we we are in very interesting times. And, uh, is that a blessing or a curse? I think depends on one's point of view. Definitely, the changes that are happening have already happened. The changes that are happening and the shift that's going to happen next when things are gonna go 2 x, 5 x, 10 x from here, things are gonna be magical. So don't get into doomerism. Don't get into thinking that the stuff is not gonna work. Don't get into thinking that this stuff is not useful, and don't get into all of that stuff. Assume that it just like mobile phones, just like computers, just like the Internet, it's gonna get 10 x better from here. It's gonna get a 100 x better from here. It's gonna get a 1,000 x better from here. Uh, over whatever time frame over the next few years, assume that a lot of human problems, humanity's problems can be solved given there is enough intelligence, and assume that there is gonna be a lot of intelligence on tap for cheap just like the way solar is driving prices to nearly zero. AI availability is going to be driving prices through competition through nearly 0 for a huge amount of intelligence on that. Do I have the intelligence to be able to near infinite intelligence on tap? I don't know. But that's the journey that we are all on in the next, uh, few years to figure out how intelligence much greater than any of our individual intelligences, much much greater than that on tap. How is that gonna change the world for the better is the question of our Lucknow.

[00:35:20] Rajan:  Types. Prasanna, let's wrap up this episode here. I know we spent close to an hour maybe, and it did not look like, you know, we spoke for that long, but always great when we chat about things like this.

[00:35:31] Prasanna:  Yeah. My pleasure, man. I think we should do more of these.

[00:35:34] Rajan:  Yeah. Let's set those up. Alright. That's it for this episode of pivotal clarity. This is an Opeka podcast. Opeka is an accelerator for global Indian founders building AI software companies. We're exploring the fast changing world of AI together with our listeners. If you like this podcast, you can find more on our website and other popular podcast apps. Subscribe if you want to keep up.

‍