India GameChanger was joined by Aashish Mehta, the CEO at nRoad. nRoad is a culmination of many years of the need to address an intractable problem in the enterprise – extracting, understanding, and providing insights from unstructured data and documents.
Some of the main topics that Aashish discussed:
- Focus on the value, not on the valuation
- AI is not a magic wand that instantly solves all problems
- The power of unstructured content processing
- Data privacy and highlighted the importance of protecting sensitive information
- Living in the golden age of technology
Some other titles we considered for this episode:
- Anything Is Possible as Long as You Want It
- Patience Is Very Very Important with AI
- When You Automate, You Are Keeping Your Institutional Knowledge
- AI's Impact on the Workforce: Embracing Change and Maximizing Potential
- Overcoming Fear: Embracing the Potential of AI
Read the best-effort transcript below (This technology is still not as good as they say it is…):
And we are good to go. Hi, this is Michael Waitze and welcome back to India GameChanger. We are joined today by Aashish Mehta, the CEO at nRoad. It is so great to have you on the show. Hopefully it's not too cold this morning. Anyway, before we get into the main part of the conversation, why don't you give our listeners a little bit of your background?
Aashish Mehta 0:23
Oh perfect, Michael. First of all, thank you. Thank you for having me. Thank you, for your listeners for actually indulging in me. You know, before I started in road, you know, I'm one of those guys where we had three other successful exits, built three companies, this is not my first rodeo. I started my entrepreneurial career back in 2005, and six, when I became a part of a group of folks, we created this company called Corporate fundamentals sold out in 18 months of existence, given that given we were so disruptive in terms of what we did, obviously, all my all my ventures have been around around finance and technology. The second venture was a company called rage frameworks, phenomenal run tenures, lots of revenue, lots of people had a lot of fun with that. Great exit back in 2017. Oh, yeah. And it was fun. And then obviously, I had to work for a quieter culture impact of yours, you know, one of those morado guys, again, entrepreneurial each is that that doesn't go away, wanted to create, build something new, build those, build that revenue, I love that go from zero revenue to whatever millions of dollars. That's what I really love it. So I'm back into bank into chase of revenue and building and hopefully really creating some more differentiated value.
So a lot of things to unpack there. I'm super curious about the difference experience of difference in your experience between the first business and the second business. If your first venture kind of gets bought at 18 months in, at some level, and again, these are all relative terms, right. But at some level, it must feel like easy. Do you know what I mean? And I don't mean, like, none of it's easy, for sure. Right? Maybe you can walk me through that a little bit. Just like what that experience felt like, so people can understand what that entrepreneurial feeling is. But the next thing is, and don't forget this either. Is that your next company? We're there for 10 years. Right? So it's completely different. And I'm curious there as well, if during the second company, we're like, we're 18 months in isn't this thing over yet kind of feeling? And then three years, five years, you're like, Dude, I mean, I will understand how those things are different. But first of all, what's the feeling with the first one? Oh, it
Aashish Mehta 2:46
was actually it a feeling of actually fulfillment on time, but quite frankly, it's like, you've you felt like you created something that somebody wanted it? Yeah. Which is incredible feeling. And then when that thing happens, you feel like Ah, okay, you know, we created value, which is super, which is actually super fulfilling experiences is one of those things where you, you have almost experience it to actually really feel how it made. You go home, and you actually go, after all this is done signing is done, papers are shared and whatnot, it go home, and you get into bed and you think, huh, that was surreal. That's exactly right. But, you know, what are the things I want to stress on? You know, I'm sure your listeners, some of them are young, you know, if they're young entrepreneurs, and middle folks will want to get into the entrepreneurship. This is not you know, none of this is actually our plan, right? These are the like series of accidents that may happen. But as long as you focus on the value, and what is which is what my mantra workflow, mantra has been focus on the value, everything will sort out by itself, don't play the valuation game valuation will come to you. So 10 years, that was a long haul. But again, it was the right haul. And you know, we'll build a building basically a value that was significant. It's a It's endurance game, my friend, you got to you got to stay in there. And you got to motivate yourself every day.
Do you sometimes watch what's happening in the startup ecosystem globally? And think this idea that you just said, I love this statement, focus on the value not on the valuation, and feel like sometimes there's been this ethos that's been pumped into the startup community that says your valuation is actually more important than your value add. And someone like you and your co founders, and your co builders must look at it and think, you know what, while they're chasing valuation, we're chasing value and the likelihood of success for us is so much higher than likelihood of success for them because they're chasing the wrong thing. Does that make sense?
Aashish Mehta 4:57
That's it that's it. Listen, I think you know, I have been around In this blog, I have been around again, I have seen few ups and downs and I have seen the.com.com Boom, the bust, I have seen the 2008 financial crisis, you know, if you agree gray hair you look great, by the way so, so you kind of learned from those mistakes. So, and again, learn from some of these, some of the just experiences that devotee have gone through and focus on value, generating real value, real cash, real cash flow. It is so important. And as I said dilution I've seen last two years. Michael, let's go back last two years, see what has happened. Two years, the valuations was way high. We know we all know, everybody was like, you see LinkedIn, you see, Twitter, you see, open any financial newspaper, oh, my God, you know, this company has actually raised $100 million $200 million 500, the money was just flowing. No, money was flowing. Money was cheap, worldwide, everywhere. The cheap, we're talking about 0% interest rates, right? When it was due in the value chain. Well, guess what is happening now, same valuation that the company had to grow into that valuation. If somebody values you, at $100 million, you have to grow into it. This is a that's a huge statement. And if you don't guess what the founders actually do, Lucia, I want
to talk to you about this specifically, because you said you've been involved in finance and technology your whole life. I don't know if I told you this when we were prepping. But like I actually sat on a US government trading desk, I used to trade US government bonds. And in a really simplistic way, you know, that, like interest rates, or inverse have an inverse relationship to price. And I think it's the same thing in the startup world. When interest rates are low, the prices of startups are naturally going to be hard because money is free. And when money is free, you spend it on things you necessarily should not spend it on. Were you aware, were you paying attention when interest rates when the Fed when the Fed started raising rates, right, because it went from zero, essentially, to what like 5%, or five and a quarter percent, this is going to have terrible consequences for people that are not just borrowing money, but that are raising money, right? Because that's the cost of money goes higher, the cost of using that money also goes higher. Were you paying attention to this and watching what's happening in the startup market as well.
Aashish Mehta 7:31
100% 100 was and so I've been actually also fortunate to be lp in one of the venture firms called differential ventures, we look at some of these things, as well. In fact, Michael, I had an opportunity to raise a bunch of capital back in when the valuations were very, very high. And from from a venture fund, and they were so happy to actually give me the valuation that I refused to take. Right people actually, you know, when I tell the story, people, what do you say? Are you nuts? Why didn't you take that kind of valuation? Right? You don't get it? That means I have to grow into it. Exactly. Don't I have to if I, if I, you know, and once you actually raise money, you got to continue to raise money. That's a you know, we gotta anyway, so I think that kind of answered the question, what kind of answered a couple other questions as well along the way? Sorry,
no, I think it's really important for people to point out I mean, look, we want to have this conversation. So people can learn something from experienced entrepreneurs about what it really means to be an entrepreneur. And also to pay attention to the things that are really important. And I'm just want to go back to this thing, you said value, what value is adding value is more important than creating valuation. And you're right, and I just want to make this fine point. And then I want to get into a little bit more of your background before we get to end road. But if you raise money, I'm going to pick a random number at $100 million valuation, you better justified because the next time you go to raise money, you can't have a down round. And no one wants to hear the excuse of rising interest rates. So you just have to you have to build into it. I'd rather have a $50 million valuation or 40, or $25 million valuation and build into 100 Later, rather than start with 100. Now because it's not about my ego, it's about building value. Yeah.
Aashish Mehta 9:16
I decided to sort of write it is it has become an ego game, it has become a social media game. And I think it needs to actually entrepreneurs who will not who refuse to pay attention to these details will be paid, there will be consequences, and unfortunately and other consequences, or they'll lose control of the company. Yeah. There's always a preference. From a money man. Money Man always has a preference. Like we got to remember that. Yeah.
So this is another thing that people don't talk about a lot. It's like once you take on a venture capitalist or frankly any other LP Right, I mean, any other investor right? You at some level, you have to answer to them, right because they have an investment framework, but they also have investment time. frame. And if they're thinking I'm gonna get returns in five years, as you get closer and closer to that five year return profile, something's gonna give. If it's seven years, maybe it's better if it's seven with 10. Like, the structure of the fund matters, too. Right? Yeah. Like the end game is not raising money. I'm just gonna go back. And re the end game is creating value for yourself and for your customers. And for the people that use your stuff. How did you get involved in finance?
Aashish Mehta 10:29
Oh, you know, so long story. Tell me I'm Lou is I actually grew up in India, okay. Well, I'm an engineer out of India. I've been to I've been to a, you know, very good engineering school. I wanted to actually be an engineer, right. And, you know, build some stuff, but not so I became, you know, I was a mechanical engineer by trade by Ray, mechanical, and industrial engineer. Then I realized, you know, what it is really random thing happened the last or the third semester of the third year of my engineering. Somebody left a book on my, in my room, and I think they forgot it. From, it's actually the autobiography of Mr. Lee Iacocca,
Michael Waitze 11:20
I love it, that it,
Aashish Mehta 11:22
I put it down, I put what actually really, you know, and I'm getting to the finance part, couple seconds, you know, allow me to allow me to boot into Waitze. And boot was fascinating. The guy just comes out of, you know, Eastern Europe, goes into all the whatever you call it, you know, the old Soviet whatever he comes, he goes to America, he is the father of my most favorite car called Mustang. Right? Mustang. I didn't even know he created it, right. But I knew Mustang. But he creates that. And then how he actually transformed from becoming an engineer to a business person, and gets into home finance, part of it gets into whole sales. Part of it, how he actually marketing part of it kind of fascinated me. So two things that took away from that book, hey, anything's possible, right? As long as you want it, yeah. And be, you know, maybe there's more, there's more than in, you know, engineering is great. But if I can actually learn a couple other things about finance, and you know, the marketing and the sales part of it, I may be, I'm able to actually create even bigger value. So I need to be able to sell. So I think there's a one line in this as in that book is as you must be able to sell what you what you create, if you cannot sell what you create. That means you're not that it is that other is a product market. fitment is problem or there is a problem with you in terms of, you know, your sales ability. So
the story of the development of the Mustang is actually really, really interesting. On top of the story of Lee Iacocca, because of all of the internal arguments that had to take place for him simply to be able to do that. There was no history in the United States of building that type of sports car. That was all European, and right for that guy to actually be able to accomplish. That was amazing. And actually, if you go further on, I don't know when that book came out. But you have to understand like I grew up in the 70s and 80s. So I literally watched the transformation of the car industry from the big three in the US. What was it Chrysler, General Motors and Ford, just getting killed by the Japanese Chrysler itself almost went bankrupt. So Lee Iacocca literally left, you know, this guy left Ford to go to Chrysler and the entire world thought he was insane. You're gonna go make my vans, like, really, that's what you're gonna do. But there's a big learning here, I think, for people to understand. And you're right, is that you can do anything if you really put your mind to what he turned Chrysler around and turn it into a profitable company and actually paid back the government. But more importantly, he understood the way the financial side of that business worked and revamped the entire finances of Chrysler as well. And that was game changing.
Aashish Mehta 14:13
I thought I was game changing for sure. And then that's kind of you know, that was the beginning. I'm like, Okay, I gotta understand what this whole finance is all about. Yeah, right. You know, I grew up in a family that is actually a business family. The finance was built into it, like, you know, we produce things and we create things and whatnot, you know, painting and so everything is about ROI and everything is about Cuban and all kinds of stuff IKEA all that I put here, you know, this is something that I've been hearing, but now I needed to get into the Finance Finance and that's what I decided to do. So that's my theory about finance. It's kind of a long ended but you know,
the reason why I like to tell these stories so she's just so you know, is because anybody who's intelligent and educated can be good at anything they try look to be great at it, you really have to care about it. And in some cases, there needs to be a trigger for that caring. And whether that trigger was necessarily the conversations you heard around the dinner table with your family about business and finance, which was just like embedded into your life, or the actual reading of the book about Iacocca, in a way, it doesn't matter. But there's something there. And that motivates people like you to be better and to be great. And that's what's really cool to me. I want to talk about Android. Talk to me about what it is how it was started, what the ideas, all of it.
Aashish Mehta 15:32
So Andrew, this has been my passion, I would say unstructured content processing, what Andrew does is a pure play on structure, content processing play, right. That's all we do. Now unstructured content, Michael has a massive definition to it. Like, the scope is so huge. So you were talking about anything from text, do videos to audios or whatnot, right? There's a there's a, there's a whole spectrum of it, we are obviously focused on the text and documents are afraid our hashtag is pretty wild. And it is actually called hashtag warrant documents. And again, not very, not very pleasant, you know, especially in the current environment, and so on, so forth. But we really mean it, we are going after there are tons of documents that are actually in the system today. Like as again, you might know unstructured content, we are producing so much angst on social content today. Like everything that you see on social media, everything that is actually out there in the world wide web, if that needs to be analyzed, that is all on structure that needs to pay, if that needs to be analyzed, you need mechanisms to actually analyze it. And again, the idea of intro quite frankly, started back in 2005. And six, when I created when we created a company called Corporate fundamentals were in we were the industry first in terms of processing, the 10 ks and the Q's, you know, you remember those Edgar filings they call it at that point, state of the art in the industry was dual screens, people were typing as fast as they can, they're looking at one screen and typing in another screen. Obviously, after after that revolution, the jobs are ready to actually move overseas to actually cut the cost of data processing and all kinds of stuff that happened but again, here we were this six people company seven people who company which basically threatened to change the entire paradigm by basically saying that I can I can provide it why why do you why do you just like typing this information, the weather in the United States, or you're typing the information in Bangalore, you know, it doesn't matter, just automate it. So we were very disruptive. But what I saw then was this massive opportunity that is actually going to come up during, in, in as regards to unstructured content processing. So kind of stayed with it, continue to research it on in my, in my spare time. I actually, in good faith, I tried to actually make this thing work with, with our acquirers, you know, did not really work out they had different, like different, you know, way of working, you know, like working their panels, and so on, so forth. And so decided to actually start it.
You had this idea in 2005. But the world of technology back then was just so different than it is today. It doesn't feel like it moves that fast. It's like, what's that famous phrase about bankruptcy? It happens really slowly. And then it happens immediately. And I feel like sometimes I feel like sometimes in the tech world, it's the same thing. There's these like little incremental changes, and then you wake up one day, and the entire world is different. I'm just curious, like, what had to change or what things had to change from a technological perspective, to make the idea that you had in 2005, build on the sort of research and work you did in between the time that happened? And enrolled, right? Was it throughput? Was it connectivity? Was it compute? Was it the combination of those things like what had to change?
Aashish Mehta 18:53
I think I believe it's a combination. And quite frankly, the industry has evolved. As I said, my first attempt to solve this problem was back in 2005, and six, we actually apply when we applied the, with, like, initially, our we create a Java based parser to extract data from these unstructured documents. You know, that sounds very archaic. Like, it's like a Fortran program. We actually created that, you know, it solved the problem. But it couldn't scale. We knew it would wouldn't scale, because it had limitations. There was no AI background, there was no period. Then what happened? You can you see that, you know, and oh, by the way, Michael, one more thing, when we started to actually store this capability to our, you know, prospects or slash clients, whatnot. They almost thought that we are playing some sort of a Jamaican who
has held rallies. So this is what I wanted to ask you, though, when you do go and show them right, because unstructured data just by definition is really hard to manage. It's in a way it's like herding cats. You know, Do you want to do it? You know, you want to get it all organized, but it's unstructured bi, it's in the it's in the definition. So you have to do a lot of work on that data. I'm presuming at the beginning, just to get it into a format that you can actually do analysis on it. I'm super curious about how that works. But when you do go show people, like, here's all this unstructured data, we're gonna drop it down to the bottom of some metaphorical funnel, and come up with some insights. Again, it must look like magic. I will call magic you call it voodoo. But it does look weird. Right?
Aashish Mehta 20:27
Exactly. And that's exactly what happened back in 2005, and six, so the maturity that now the maturity of the, you know, the buyers have actually exponentially increased the like, like, if you will fast forward between, from 2006 to 2023, or when we were looking at completely different different different actually gonna be game, it's changed the world, right? The computers come down, there is a massive amount of actually massive blue, a massive amount of AI tools that are actually available that have worked, okay. We have emergence of cloud players, we have, you know, we have emergence of AI technologies, AI technology has gotten better language models have gotten better. Now, we have now fully charged GPT we have alarms. So there is a dis problems. Now, the given all this tools that are out there, it is now becoming, that was a different problem. Not everybody thinks they can solve solve. Like, yeah, you know, what do you do? But I can I can apply, you know, chat GPD to do this, like, No, you can't, okay, but you know, give it a shot. Now you give it right. So it is a complex problem, as you said the Data Prep aspect of it, before you will actually go to the insights part of it. It is one of the hardest problems that he does all. Like, think about, right? You have a document let's say you have you know, you're a bond trader, you get a Bronco bond prospectors, let's say you, you know, which has like, I don't know, 500 tables. financials are coming up 500 times about the prospectors now expenses actually in that, right? There are a bunch of text, there is a connection, you know, there are footnotes that are notes within the notes, there are tables. Now, if you have this, and you basically say hey, and road, analyze this for me, think about the how complex that is, first of all, you knew I did, the tables are utilized. Right now we're talking about technology, not on what your eyes, you know, human mind, you're talking about technology. Now you're identifying the table. Now you're now you're basically saying, How do I create the name value pair from this table? What if there is a table within the table, there are nested tables, when there are tables, sometimes there are borderless tables, you know, people get crazy with their, with their imagination in terms of how they represent information. That was some sometimes tables can actually go from one page to another without any kind of reference, right? That tables in dual languages, like especially that's very prevalent actually. Around the World. Yeah. So these things, you have to be able to flatten that create the name value pair so that you can analyze it. Think about it. There are multiple row headers, and there are multiple column headers. Yeah, it didn't one column, your three columns, you have all within one row that will have multiple rows, right? So one, one could say 99. If like yours, somebody said that there is a division, then say gender, whatever it is, right? You have to be able to actually analyze all this. And I don't mind the details. And if you don't, then your your your resultant insights will be actually inaccurate.
Is there a real time? What's the right word? Is there a real time factor to this as well? How fast is it possible to process all this unstructured data? In the sense that I think about it as from a real time trading perspective. And here's what I'm thinking about. I'm looking at like a virtual ticker tape, right? So all this information on stocks, let's say everything that's in the Nikkei 225. But while I'm watching that trade, I'm also watching all the news around the Nikkei 225. But then I'm also watching all the news on the foreign exchange. So these are all three things and then the interest rates, right, because all four of those things react with each other. Right? The tech fast enough yet to be able to analyze that unstructured because it's all unstructured, right? It's coming in, in particularly the news in different formats. Is it possible to do that? Or are we still in a situation now where we where it's not fast enough yet, but we will get there?
Aashish Mehta 24:35
We are actually getting there. I don't know about that yet. In many aspects. And then there are a couple of there are also a financial angle here. You know, what is the real value? You know, because sometimes when you need stuff real time, are you willing to pay for it? Because see, if you're making a multimillion dollar decision, and if you think there's an implication there obviously right you can you can you can be Sleep pay for the compute reward not because real time processing of such information has a has a cost of some massive cost. Yeah. Right. Now, having said that cost is one aspect capability wise, are we there yet, if you we are sufficiently trained, we are there like today, Michael, with some of our customers, we are actually getting over a 50% split to accuracy, what that means is you upload a financial document into enrolled and 50% of the time it is going straight through, like it's just flying through the system. So when you click it, on the other end, you get the stuff you need, the variables that you need 50% of the time, right, so and so on. Now, the reason we actually have achieved these for these customers, is because we took sufficient time to cleaning it, making sure and there is also a there's also a third factor, which is really the as I said, What is the training times second is the cost implication. Third is you got to be patient with AI is the AI is not a magic and AI is not phenomenal can can think on its own, you know, I don't believe in all that right, you know, you got to teach AI. So think about it, let's say you start actually working with the with our model, or any model MRO you're teaching the EMR what to do, then the AI model is making mistakes, then you are correcting those mistakes, and then you're letting it fly again, then you're correcting it again. So over the period of time, we know we got to give about three quarters to really set the whole thing down, and then only you can actually see some of these. So patience is very, very important. With AI. And you will get it you will get it as well. As you know, what has happened also is right software buyers, called bison, John Doe, are the consumers of software, there is almost instant gratification feeling oh, I click the button it has to work. So that is right. That's absolutely true. Because and I don't blame anybody it because so far we have solved deterministic problems. Now we are into the the problem, we now we are into probabilistic problems that we are solving. And probability by definition is not 100%. Right? So it's got to be at all, it's not deterministic. So we got to be patient about that.
More people should understand advanced mathematics and at least calculus so they can understand iterative functions and understand that when you're trying to find a limit, you're never going to find it on the first try. It may feel like you do, but you don't. Anyway, a lot of people are gonna have a hard time processing that I want to ask you about this. People think a lot about natural natural language processing, when they think about how AI goes through data and text, I understand what it's about, I want to try to understand, if you're taking all this information, you're analyzing all this data from unstructured data, right? You want to get it into a place where it's it feels like it's natural language, genitive is generative as well. So that what the output is you people can understand how hard is that part of this
Aashish Mehta 28:09
god part is there, like now the generative AI models are actually getting better. You know, some of the social, I think incredible research that Obama has done, and Google and Facebook, Facebook models are great. Some of these models, and then there are tons of other models that are out there. It's all it's all there until you the trick here is in my view, it is for especially I'm talking to our enterprise clients, our enterprise clients, they have a they have a bigger issue here, which is how much data they're willing to share with these third party providers. Data cannot go out the all these are federated models, or these are actually reinforced threaded and models, that means that you have to like when you're typing, you go to Bard, or you go to Open AI today, and you start actually testing GPT, whatever you're asking, or whatever you are actually teaching that goes through the main model, which is owned by, right. So you are actually willing to give that information away to the world, right? As a consumer, you can afford to know because you don't really care, you know, maybe you don't care or you care. But as a as a as an enterprise, you're heavily regulated, you need to be actually protecting your trade secrets and information that you will generate. This is where it gets complicated. So it is models or they're adding the how you implement it, how do you actually apply the internal knowledge is actually going to be the key.
So I think this idea of like taking private company information than necessarily should not be shared to the public and putting them into just a readily available LLM. Right, or even a calculation engine that's available available publicly. You're not giving away proprietary data. So there's a necessity to be Old private models. I'm curious of those private models haven't been trained with the same depth and breadth of data that the public models have. Is that fair? I'm guessing. So how do you? How do you do that, though? Because then you're talking about just this very specific data that gets that gets trained? Or can you get access to the public data, bring that data inside your own private model? And then run your own analysis with your own data based on that data as well? Does that make sense?
Aashish Mehta 30:26
Yes, absolutely. Anything you say, right. So basically, these are we call, there's a whole process of quantization that needs to happen. You take the model, take the mod, and then quantize it, quantization is really your, you're making it smaller. Like if it is actually taking the billions of whatever it is the inputs, it actually takes small inputs, but then you got to train it on the internal data, and you got to get bring that efficacy actually solve the problem like that. Interface is not solving for what is the best coffee in let's say, you are in Bangkok, right? Or wherever you are, with the best coffee and let's say Boston, and this MLMs are solving for everything, that's all important. What is the best coffee in Boston to, you know, give me the best cancer research, you know, output one, all right, and everything in between. So it was not falling for that that's actually very consumer oriented application, enterprise a solving for the narrow functions like, you know, within accounts payable, they may be solving for one, accounts receivables, they'll be solving for one order management that may be solving for one specific outcome, I think you have an ability to take these models and really drive a business outcome versus trying to actually, you know, the types of like, bunch of stuff. And that's what I that's what I mean by I think there's an opportunity here, quite frankly, you know, but the technology is there.
Are there product development implications here. I mean, it's that what you're suggesting, in the sense that, let's just talk about insurance a lot, right? So insurance is obviously a subset of finance, it's just a part of it. But a very large part of insurance happens to be a very large business. But is there a way to take all the unstructured data around, like the things that people make claims for the incidents that they have that cause those claims, the actuarial math around calculating what the risk and return is, and then maybe some new sort of risk management data put together to then come up with some very specific products for very targeted market segments as well? Is that part of the thing that enrolled is trying to solve to what their partners?
Aashish Mehta 32:31
Absolutely, that's exactly, I think you I think you articulated very well, in fact. So when you talk about insurance, insurance is a huge opportunity for a lot, right. Right now we are focused on is analysis of financial statements, we have done it very successfully, we have we have at the level where we can actually get a financial statement, we can analyze the financial statement on behalf of our clients, and then provide them variables and the signals that they need within, we have been able to successfully reduce in one instance, we've been able to successfully reduce their their size of their operations and sell them scale by or, or I would say about 140%, it's vulnerable, the time to market has, has actually improved quite a bit. But I'll tell you, the biggest advantage, while it is a soft advantage, but now I think there's a realization coming up is that when you automate using AI and models or whatever, right, you know, enrolled or not adored, what, it doesn't matter. We automate, you are keeping that interest, institutional knowledge within your within that model. Right. So you're not dependent on Blitzer, she should not depend on Michael Michael tomorrow, leaves the company, whatnot, he's not he or she is not able to take knowledge with him, you know, the model retains that particular knowledge. And that is huge for these guys now. And they have so far suffered in quest of, let's say, achieving the quarterly results and reduce cost and so on so forth. They have you know, they have resorted to outsourcing and all kinds of stuff. But I think there's a realization now that, you know, the some of these processes have to be that technology is there there is there's enough horsepower that exists today that these functions have to be properly administered.
Yeah, I mean, I think the implications alone for just this idea of that the knowledge doesn't leave the company, right. I mean, in the old days, two years ago, if a guy or gal who has an expertise just switches, locations, the stuff that they build, sometimes it'll still work, but nobody will understand why. And if it breaks, then there are problems but if the institutional knowledge is saved, basically inside the model, the implications for that are huge, I would think. Okay, let me ask you this before I let you go unless there are more things you want to talk about, but I feel like I can keep you going on this. Great, okay. I can keep going. What excites you, though about the future? And specifically, maybe just about expanding into sort of other verticals. But what excites you about this? Now?
Aashish Mehta 35:10
They'll tell you that technology is you. We're living in the golden age of technology, it feels like it, doesn't it? Oh, my God, this is actually it's scary at the same time, it's actually pretty exciting. Yeah, we Deckard, we will not be able to do not even six months back, now we're able to do. So it brings its own challenges Michael was at the same time, it actually has a, you know, I'm generally you know, as a as a consumer of technology. I'm actually very excited. As a business owner and an entrepreneur and someone who's actually trying to build a company here. In this particular space. I am excited at the same time, I don't know what what is going to come. Yeah, but I'll tell you building, building this. So I'm excited. I'm excited about see, again, we are investing in the right areas we are actually looking at from from Jenny AI perspective, we are, you know, we are investing in the right areas, we are ensuring that the alarms that are there today, can really help us actually increase the efficacy of ours, our goal is to actually solve a problem end to end and we're going to stick to that we are not going to actually just give you a model single Okay? And are you to do it yourself, and then hope that we will stick to that solid problem, create a real value, again, go back to basics, create a real value for client, which I told you, I just explained to you, I just I think give a couple of examples. We are at a minimum, we are providing all our clients, and we're so lucky to be able to do that. Minimum 50%, productivity improvements minimum wage gap. In fact, at this particular stage, we actually 24 And I only started doing it, I'm guaranteeing my contracts or guarantee. I'm like, Okay, if I don't get paid if you don't if I don't deliver that. But if I do deliver it, you got to pay me more.
When you're installing magic for people, it's pretty easy to be confident about it. Can I ask you this too? Before I let you go? Sorry, one more thing. There is a little bit of fear, right? Like there's this gigantic excitement. And for guys like you and I, we've seen a few cycles, let's just say that. And some of those cycles are small and relatively significant. But some of these cycles are really big and are inflection points. And I would make the case that every inflection point has some kind of fear attached to it, particularly for people that don't understand it. And I want to get your perspective on this. While there is excitement around how fast and how amazing technology's advancing, do you think that the fear itself is endemic, meaning it's always going to be there when there are new technologies coming particularly when they're this significant? And yet, in most cases, there, it's kind of overblown?
Aashish Mehta 37:48
It is, you know, again, some of it is completely overblown. Listen, I'm actually I'm, you know, I'm around AI systems, I've been around, you know, creating one using one or whatnot in some various capacities for several years. AI is not scary. But I'll tell you what AI is. AI is scary for people who don't want to work. It's scary, but want to use their brains AI is scary for people who want to actually do mundane task. Ai should be scary for people who actually are trying to compete with machines. As humans, you know, you're gifted, we got to be able to actually do more and more stuff. But AI otherwise is actually you know, is going to be a game changer. It is it is you like it or not, it's here. And so let's not be fearful about it. Let's actually, let's figure it out what we can do.
So you live in Boston, I want to share something with you. In 1978, I was 13 years old. And you can go back and do some research on this in 1978, there was this thing called the blizzard of 1978. And they closed the roads in Connecticut where I was living at the time, for like a week or maybe 10 days. I can't remember the exact numbers A while ago, and Boston just got crushed with snow. And I would imagine you'll see where this is going in a second though, I would imagine that the guys that were selling shovels were really mad at the guys that were selling snowblowers because the snowblowers made moving the snow kind of easy. And the people that didn't want to shovel the snow were just like the people that wanted to shovel snow because that was where their living was getting made was mad at the snowblower community because they're taking away their job. And to me, that is equivalent at some level to AI. Right? I use it every single day in my work. It's made me way more than 50% more productive. And yet the output is just as good if not better, because I know how to use it in the same way that like a snowblower could cut your foot off if you don't know how to use it. So it's scary because you got these spinning blades. But the reality is that if you know how to use it, you're only outside for like 40 minutes and it's not that hard to do. Is that a fair analogy? At least at some level? Absolutely.
Aashish Mehta 39:53
Absolutely. Listen, I think we you know, you're absolutely right. I think the analogy is very fair. See every every technical level Listen, that has happened. There was fear associated, as you rightfully said, when the internet came, obviously, if somebody's right, somebody's wrong, you know, you look back. Yeah, back in, you know, 2000 we talked about, oh, my god internet's gonna take over the world. You know, they wouldn't be no government. There will be none of that happened. Crypto, there will be no, you know, there will be no, what we call the there'll be no Feds exist, you know, there'll be no central finance financial authority is that happened? No. But crypto brought something to the world that is unbelievable. In the blockchain technology is real. I think there are real time real important implications. Internet has changed the lives of people, fundamentally speaking completely know the word, people who you know, especially people who are actually living in the parts of the world or in a world where there is absolutely no connectivity, see what has happened. It has their destiny their lives, you know. So, technology is here to stay. And they know to take a different form, but it is going to stay and it's going to improve our lives.
Michael Waitze 41:06
That is the best way. Aashish Mehta, the CEO at nRoad, thank you so much for doing this. You've got to come back with us. Thank you so much again. So I really appreciate it.
Unknown Speaker 41:15
Michael, thank you. Thanks for inviting me and I'd love to be part of more. Okay, thank you.