Inspired Execution

00:00:05:06 - 00:00:30:03
Chet
Raghu, welcome to the inspired Execution mini series. You grew up in India. You spent six years with tech companies like, you know, T.I. or I should say hotties, because a lot of people may not. Texas Instruments, they all know Broadcom for sure. And then co-founded your current company, Yellow Dot II, the name I love, as you were talking before in 2016, with a vision to transform the way customers interact with brands.

00:00:30:03 - 00:00:47:08
Chet
Right. And so sounds like an awesome journey, right? I mean. Tell us tell us about a couple of magic moments. You've had three a journey, right? Right from And you can go and you can go to your early part or if you just want to talk about your magic moment at at Yellow Dot II, that would be fine as well.

00:00:47:12 - 00:01:22:22
Raghu
Yeah, a bit of fantastic journey, Chet, and excited to be here. A lot of my earlier part of the journey in 16 years, being walking into a mechanic to take off the most special part has been my first work experience at DIY. I mean, you know, it's unfortunate that it's not out there, but then I was just out of the college that was probably the company to work for developing chips, for Nokia, etc. I think the fundamentals are being about how great products and how great tech culture can be.

00:01:22:22 - 00:02:07:07
Raghu
Building companies that thought I'd experienced lost time, but a lot of magical moments, of course, happened during the entrepreneurial journey over the last last eight years. I would say that, number one moment that we found it super magical is we had this large financial services customer and we have a tier to product market fit and we were going in and deploying this automated shock assistant for them and get all they expected a few thousands of users to adopt it and they experimented using this for campaigns and marketing.

00:02:07:09 - 00:02:33:16
Raghu
And within a few days our servers were getting exploded because the number of users turned out to be a few million users and what was a magical moment where we felt a real, real product market fit for the company and that also gave us a lot of learning is a lot of our product development hasn't come from what we believed, but it also came from all customers fault.

00:02:33:18 - 00:02:54:10
Raghu
So one big learning is I mean, always valuing our customers versus push customers in the sense that the customers are pulling the product or fuel because they have a vision around what you are developing. There is a vision that aligns to what you want to do as a company, and they see in a lot of missing features and they gave us a lot of pain.

00:02:54:15 - 00:03:22:06
Raghu
But the product is what it is because of that. The second magical moment was being big in India, starting the company in India, expanding to Asia-Pacific. We were always looking at some time to get on to the North American market and expanding there. And this is one of our first U.S. customers, Schlumberger, where we just talked about, hey, we were using this product to enable customer support automation.

00:03:22:08 - 00:03:51:14
Raghu
And there was this pioneer, a visionary within Schlumberger, who came back and said, you know, we don't have these millions of consumers, but we have a lot of employees where they spend billions of dollars in supporting them. Can we use this product to support our employees? And that became a platform because now the product has expanded to are completely different use case again driven by vision of of one of our customers and prospects.

00:03:51:18 - 00:03:58:07
Raghu
I think these two have been you know the super magical innovation.

00:03:58:09 - 00:04:24:00
Chet
That is awesome. That is awesome. But what a what a great journey. So you don't as you talk about Schlumberger right then and you had did a two part question. One is, you know, for those people who don't know, well, how would you very crisply define conversational AI? And then the next thing is what does zero touch customer support mean?

00:04:24:02 - 00:05:02:22
Raghu
Let me get into the conversation. The AI. So the technology that orchestrates dialogs, dialog between a human and a machine is essentially conversational AI enterprise. Conversational AI is back narrowed and focused towards conversations about an enterprise. That's what their customers and implies. So that's that's the broad definition. And, you know, there are lot lot more details on but it is conversations also get past Don or they're just, you know, conversing about, you know, what is the impact that they use and announced that they can use machine learning.

00:05:03:02 - 00:05:19:10
Raghu
But the technology keeps advancing. But at the core of it, any technology that can orchestrate dialog between a machine and a human is conversational. AI. I think the second question has been Sorry I missed that. Can you repeat that.

00:05:19:12 - 00:05:22:05
Chet
Question was nearer touch your touch customers.

00:05:22:06 - 00:05:25:12
Raghu
Yeah. So what we think with that, what.

00:05:25:12 - 00:05:27:19
Chet
Does it look like? What would be a before and after?

00:05:27:23 - 00:05:50:13
Raghu
So I kind of take the analogy of self-driving cars. Chip Right. So it all started off, you see the evolution of self-driving. It all started off with cruise control. Hey, I can I just go in this lane then? Now you have no lane assist and you know, you see the evolution of Tesla and.

00:05:50:15 - 00:05:52:02
Chet
It's just driving things like.

00:05:52:02 - 00:06:02:04
Raghu
Yeah, exactly. And the vision for self-driving is that he, of course Cruise got into some problems and you know but but essentially.

00:06:02:06 - 00:06:04:21
Chet
The plan has not completely shipped and that's.

00:06:05:02 - 00:06:34:13
Raghu
What FSD is the vision valve there is almost zero human pouch in driving the car to complex and out of roads and city traffic. And similar is the vision that we have for customer support there. I think the technology award, from providing answers to enabling path and we believe and the company is going behind the mission that customer support one day did not have any human body to answering questions.

00:06:34:13 - 00:06:51:15
Raghu
The probably managing the system by they entire customer support can be absolutely zero touch just as a full self-driving is far for commute and driving. So that's the vision behind the zero touch customer support.

00:06:51:17 - 00:07:16:19
Chet
And do you so this I'm so let's actually talk about this you started yellow dot AI before the Jenny AI wave yes right and how is that been have you had to pivot a lot of technology stuff did you you know when we when the book gets written we see AI part about it in 2016. And you know what it was great with the but that's generally not the case, right?

00:07:16:19 - 00:07:26:14
Chet
Because you make it look like there's a, you know, a PC on every desktop, right. But but what was it? I mean, was has it been very hard shift or is it been a small shift for you?

00:07:26:16 - 00:07:51:08
Raghu
So the technology has been ah, a hard shift for sure and hard. And I mean they are excited the way that it's changing. But I think the problem and how it's going to operate and thus stayed constant. So even that and when we started the company, we completely believed that interactions between enterprises and their customers and employees are going to be automated.

00:07:51:08 - 00:08:19:13
Raghu
And we are a company that are going to help enterprises do that. And it started off with machine learning information, providing information, then it more transactions and then it more to live it off. I mean, that last language models in 2018, 2019, which are not as large as they are, but the models are blackboard, you know, reasonably large at the time and not being driven by an allowance.

00:08:19:13 - 00:08:39:17
Raghu
So the technology pleasantly has been, you know, it's been moving very fast and that's helping us moving faster to provide more delightful than what I call the experience, as do our customers. But the core of what we want to do, the business model there remained constant.

00:08:39:19 - 00:09:05:00
Chet
That's awesome. So as you think about Jenny, I right, I've always and I want to talk about technology in a second. Go back to it. You know, I hope I get a chance to talk to quite a few CEOs every week and CTOs and now, by the way, CEOs as well. Just because Jenny is on everybody's mind. And I always talk about the incremental use cases and the transformative use cases.

00:09:05:02 - 00:09:35:01
Chet
Right. And the incremental use cases are doing a lot of what, for example, you have been doing. Right. Which is I'm going to do zero touch customer support. You want to use it internally. Let's do that. Like Schlumberger, that's a good example. And you know, people are doing customer facing things. The second the transformative stuff is when you start thinking about not just chat bots, but you think about autonomous agents and you are coming up with things that change the panel and you're coming up with new revenue models.

00:09:35:03 - 00:09:55:06
Chet
So do you do you agree with that? Because this is generally how the waves have gone, right? If you think about it on the Web, you started with, you know, on the phone, on the on mobile, you started with you know, you you started with Angry Birds and then came your banking app, Right? And the banking app changed your life because you didn't go to a retail bank anymore.

00:09:55:08 - 00:10:04:19
Chet
You didn't go to a branch, Right. Do you agree with that? You know, classification of how genie is going to play out from an incremental and a transformative, transformative point of view?

00:10:04:19 - 00:10:37:14
Raghu
I think there's a similar topic that I was discussing on on another forum, and here is why I believe generative AI is going to play completely different from mobile and cloud. The number one reason is if you take even mobile or the cloud revolution, the core functionality of for the end user remained constant. For example, if you take a CRM or a ERP, you know, basic thing of applying leaves or getting customer information, I'm at a broad level.

00:10:37:14 - 00:11:07:04
Raghu
Of course they do phases and you know, workflows changed a little bit. But essentially what Salesforce brought to the table was, Hey, this all is delivered from single multitalented model that can be reused versus on competition, but the functionality has not been a certain exchange or a fundamental change similar with mobile. I think the functionality that the banking application would have had on the web, you know, transactions, etc., they continued out.

00:11:07:05 - 00:11:34:06
Raghu
They moved to a different surface, which is mobile and and brought in the functionality. But what Ginny has brought to the table is the core change in the functionality itself. And I support changing the functionality. What is it you're now able to generate content you are able to generate completely new videos. It was never possible. Now, Virginia, you can enable software to make decisions.

00:11:34:08 - 00:12:21:05
Raghu
The software just from storing information to connecting and managing relationships with people. So the general change is going to be much larger, much more fundamental, much more disruptive than our cloud, our mobile teams. I would kind of related to almost like an Internet kind of change because if before Internet and after Internet, Internet fundamentally changed what people can do, you know, with the computers and this is as fundamental as that, whereas cloud and mobile are, you know, variations of the functionality delivered through our different surveys are delivered from a different infrastructure.

00:12:21:07 - 00:12:38:07
Chet
Yeah, I agree. I agree. I have a different way of saying exactly what you said, which is this is generally if you look at tech waves, right, if you look at clients where you look at mainframe, they were disruptive by themselves. They only disrupted the way before the thing about Jenny Ise, it builds on the web, it builds on mobile that builds on cloud.

00:12:38:07 - 00:12:58:20
Chet
Yeah, right. And and it's and it's the first technology probably in a very long time. Yeah. For a technology that actually behaves like human beings do. Right. And so it's a, it's a, it's a very interesting thing of how quickly it'll accelerate. So so But go ahead. You had something.

00:12:58:22 - 00:13:02:11
Raghu
I think that I completely agree with that.

00:13:02:13 - 00:13:22:19
Chet
So there's a lot of debate on, you know, are you going to have one model that rules all models and or you're going to have a large model, but you're going to have smaller specialized models as well? I certainly believe that the latter, not the former. Right. Do you what what is your take on it?

00:13:22:19 - 00:13:52:08
Raghu
Yeah, I fully agree with that. And here are my reasons is at the end of the day, the model, what the model can do is dependent on the underlying data, assuming that you have, you know, the models keeping increasing and yet all the parameters go from trillion to 10 trillion. This has become super, super smart. But at the end of the day, they wouldn't have the knowledge that is stored within the enterprise.

00:13:52:08 - 00:14:18:11
Raghu
Firewalls are enterprise systems. To give an example, just for, you know, I'll be working with one of the customers and they are using this child voice assistants to drive more conversion from their marketing users. There is no way that the NLM is going to know what kind of conversations converted, what kind of interactions didn't come word, but so all that information that is lying within the enterprise firewall.

00:14:18:12 - 00:14:50:04
Raghu
So you would have this really large foundational model that has all the world's information, but that needs to be supplemented by smart models that have knowledge of what happens behind the firewall. All behind are within the enterprise systems or wherever the data that's not accessible publicly. Right. I think a combination of this is going to deliver the best outcomes far for companies, individuals, etc..

00:14:50:06 - 00:15:14:13
Chet
So in this world, are you architecting that way as well, do you think, or do you, as you think about your platform, do you think of using a large alarm? Yeah, right. And I would like, you know, a chat up to a bard or whatever anthropic, whatever it might be. You think of that as it's not either or you in from your point of view, not from your software's point of view.

00:15:14:13 - 00:15:19:10
Chet
You think of it as Yes, do both. Right, because you will leverage those over time.

00:15:19:15 - 00:15:49:06
Raghu
It is, yes. Both from a my perspective and our software, even right now in production, uses exactly the same architecture where there is orchestration and alarm that determines what is the right model for it to call. And there are use cases where we call the formation of the models bid long auxiliary, but they're also fine tuned models for specific use cases that get called upon.

00:15:49:08 - 00:16:03:16
Raghu
So absolutely, both from my perspective and our company's perspective and our current deployments, they have this orchestration layer and it's a combination of large financial models and specific function models.

00:16:03:18 - 00:16:24:02
Chet
So let's talk a little bit about, you know, 2024 predictions, right? Share your best AI prediction for 2024.

00:16:24:04 - 00:16:53:01
Raghu
That's it. So let me let me let me come back. I think we will be definitely surprised at at the power of of Genesis are the use cases that are going to come it I think I'll my prediction is that it will be unpredictable in the sense that they use cases that we're going to see are something that our food top predicted predicted this year.

00:16:53:03 - 00:16:58:14
Raghu
My second prediction on the negative side, I think that's the positive side. On the negative side, controversial.

00:16:58:14 - 00:16:58:21
Chet
Side.

00:16:58:21 - 00:17:29:04
Raghu
Towards the other side is I think we'll see the death of a lot of general startups, 98% of it. I mean, I would say that because I think there are very few companies that are, I think have the ability to really differentiate down the foundation models and have the right business model. So there'll be that I think it'll be a tale of two cities that would be significant winners and that would be a really long tail of losers.

00:17:29:06 - 00:17:44:21
Chet
I would, you know, and Libby, double click on the on the controversial one. Do you think the you think the the startup said don't succeed? Do you think that day of reckoning comes next calendar year as in the year after?

00:17:44:23 - 00:17:47:10
Raghu
I think the next calendar year, I think.

00:17:47:12 - 00:18:04:12
Chet
Yeah, I know I was going to say the same thing because I think this because a lot of them are spending money. Yeah. Like, like people are going they're doing this, by the way, for the good for the right reasons. People figure out their staff. Yes. But the problem is companies are not going to be on companies like yours and mine.

00:18:04:12 - 00:18:23:11
Chet
We get paid on consumption. Yes. Right. Because we are cloud companies and people are not going to be consuming massive amounts until they put things in production. Absolutely right. And that is going to take a while for a company like Schlumberger or a company like we're on is in. For us, they just they they're very careful. They're regulated, things like that, right?

00:18:23:16 - 00:18:53:22
Raghu
Absolutely. Absolutely. So a lot of these applications have to get into full scale production for it to make economic sense for a lot of companies. And, you know, people are trying multiple different models. There's a bowl versus by the band. A lot of functionality is going to become subsumed into the formation models and and now you have Hyperscalers playing into this like Amazon and Microsoft.

00:18:54:00 - 00:19:18:00
Raghu
So it it does take a lot for companies to overcome the, you know, the the things around, you know, subsume functionality differentiation on top of foundation models and getting large scale production. So so yes, I would I would agree I would I would think that it's it's it's going to be a tough get.

00:19:18:02 - 00:19:30:05
Chet
That's my agree completely. All right. We're at the rapid fire stage of our off the podcast. So I'm going to go through a bunch of questions and you should try to answer them as quickly as you can.

00:19:30:07 - 00:19:30:19
Raghu
I'll pay for.

00:19:30:19 - 00:19:39:10
Chet
It. So. So what's the problem humanity is facing that you want genii to solve first.

00:19:39:12 - 00:19:46:06
Raghu
Poverty, inequity, wealth inequality.

00:19:46:08 - 00:20:11:08
Chet
What's one thing in your day to day life that you want A.I. to automate making money? I love that. That is awesome. That's awesome. You play squash on occasion. We talked about that. How many years until you'll be playing against robots?

00:20:11:10 - 00:20:13:23
Raghu
25 years at least.

00:20:14:00 - 00:20:43:07
Chet
Oh, I think it's going to happen sooner. I go be optimistic. Be optimistic. They may not be able to beat you in the next ten years, but I think we'll be playing with robots before. So at least some combination. Let's see what happens. It is a very nuance sport, right? For those those people that don't know what squash is, one word you would use to describe great tech leaders, visionary execution.

00:20:43:07 - 00:21:05:11
Chet
It's that's awesome. That's a that's a you know, it's that what I call that is range, right? You have to have you have to vision, but you have to make sure it's execute, execute. You should be able to execute the vision. Right. Otherwise, it's just a really awesome document and a good speech. One word to describe how you feel about the future of yellow.

00:21:05:11 - 00:21:10:20
Raghu
Today I got to wake up more excited.

00:21:10:22 - 00:21:20:19
Chet
Really? That is awesome. That is awesome. By the way it looks. It looks you can they can see it. I can see the energy coming out right when you say that which is awesome synergy.

00:21:20:21 - 00:21:23:03
Raghu
Absolutely, Chet.

00:21:23:05 - 00:21:30:14
Chet
Rugged. This has been phenomenal. I deeply appreciate this and hopefully we get a chance to do this again soon.

00:21:30:19 - 00:21:51:02
Raghu
Absolutely. I thoroughly enjoyed this started and yeah, I look forward to this coming up and stay in does absolutely Thank you. Thank you very much.