Full Transcript

Dhruv Sharma: Hey there and welcome to The Offline Network. If you tune in regularly, then welcome back. It's already 30th of January. This is our final stream for the month. I want to ask you again how your resolutions are holding up. Utsav, it's good to see you. I'm feeling unusually optimistic today. Do you want to know why? So the economic survey dropped yesterday. It's a document that they released just a couple of days before the annual budget. It's a government document, actually, and we have a civil service in the country called the Indian Economic Service that actually works in putting this survey together. It's always packed with insights and nuance, but this time it's also a very breezy read.

Utsav Somani: And it's got 700 pages.

Dhruv Sharma: It's 800 pages. It's got close to 20 chapters, but I've just looked at one, which is chapter 15. It's a chapter on urbanization, making Indian cities work for its citizens. And I'm actually going to quote something from the chapter, which is what has me feeling unusually optimistic today. And I quote, it says, Indian cities are aspirational, but exhausting. Isn't that remarkably candid for a government document, by the way?

Dhruv Sharma: And then it says, they demand resilience rather than reward curiosity. To become engaging, Indian cities must move from survival-oriented urbanism to possibly oriented urbanism, where public spaces, culture, mobility, and governance expand what people feel they can do with their lives. If this is the quality of thinking at the highest level of policymaking, we're likely in good hands. And maybe we have small problems here and there right now, but overall, and over the long term, things are going to be just fine. That's my takeaway.

Utsav Somani: Fingers crossed, man. I don't know how much of this is written by AI. We talk a lot about AI, so I hope it all adds up. But before that, World of AI, I mean, Google is making moves. SATs and JEEs are now, you can learn about all of those exams and self-teach yourself on Gemini. Did you give a competitive exam in your life, Dhruv?

Dhruv Sharma: I did, yes. I mean, if you hold an engineering degree, chances are that you will have taken at least a few of those. But you can take mock tests now. They can actually give you mock tests that also change. You don't have to go sit in an examination center. You can sit in the comfort of your own home. You don't have to go to a different city altogether. There were these cities in India that had become like ocean capitals.

Utsav Somani: I remember this clip that I saw on Lenny's podcast. He appeared recently and he said that the best form of education is one-on-one tutoring. And it's typically been reserved for the upper echelons of the society. And now with AI, I think curiosity is the ceiling, right? So basically, if you're excited about a thing, you can learn it yourself. So it's personal tutoring at scale one-on-one. So I think it's going to be exciting what the impact this has.

Dhruv Sharma: And there's a reason for that, by the way, which is you can't treat students as products in a factory, right? Like they all have different learning needs and different learning styles. And if you're in a given subject, if you're ahead of the class, you're going to get bored to death, you know, as waiting for the rest of it, like waiting for everyone else to catch up. And similarly, if you're relatively slow in a given subject or if you've lost references or something, then playing catch up isn't easy either. So personalized tutoring at scale is going to be, I think this is, again, putting AI to the use that you should be putting it for. Unlike, you know, some of the report writers.

Utsav Somani: But I hope there's some truth to it because some of the numbers from that report look pretty promising. GDP growth expected at 7 percent. There's a decline in inflation. There's huge capex growth, 89 percent growth in that from 5.9 to 11.2. For, I mean, 55 percent increase in the number of gig workers. We've discussed that topic previously on the show. And all of the top 10 fastest growing cities of the world are in India. Surat, Agra, Bangalore, Hyderabad, Nagpur, Rajkot, Chennai, Vijayawada. So it's a pretty exhaustive list.

Dhruv Sharma: I think Surat, you know, reminds me, so they're projecting that Surat, just that one city by 2035 could on its own have a GDP of, I think the number was $128 billion, just that one city. It's economic output.

Utsav Somani: Yeah, Gujarat in itself, I think is a fascinating case study in what they've achieved. All right. Let's welcome our first guests for the show now. Prasanna and Anand from Arya.at. Guys, welcome to the show.

Prasanna Rao - Co-founder, Arya.ag: Great to have you. Thank you for having us.

Utsav Somani: Anand, thank you so much for giving us the time. For listeners who are hearing about Arya.at for the first time, can you describe your business?

Prasanna Rao - Co-founder, Arya.ag: So we work as a platform that solves for two very basic problems in Indian agriculture, which is that farmers, especially the smallholder farmers, which constitute about 85% of all farmers in the country, do not have the choice to really decide when they want to sell their produce after harvest. Invariably, they end up selling immediately after harvest, where the prices are the lowest. So we solve for that particular problem. And the second problem that we work on is to see if we can enable choice for them on whom they want to sell that produce to. So like they do not have the choice to really decide when they want to sell their produce, when actually they go out and sell, even at those distressed prices, they end up selling to buyers who are in their immediate local network. And if they had the choice to really reach out to buyers outside of their local network, buyers, large corporates, bigger processors around the country, they could earn at least another 15 to 20% more. So our work is basically focused on enabling choice for farmers and their organizations to be able to decide when they want to sell their produce and to whom they want to sell their produce to.

Dhruv Sharma: So now we're talking about pulses, cereals or peripherals.

Utsav Somani: I want to congratulate you on the milestone first. You've hit 725 crores, your fresh raise as well.

Prasanna Rao - Co-founder, Arya.ag: That's right. Yes. Thank you so much.

Utsav Somani: This post-harvest process, why is your business uniquely or what is the source? Like why is ARIA.AG winning in this world, the post-harvest world?

Prasanna Rao - Co-founder, Arya.ag: It's still the beginning is what I would say. I mean, if you see what we do today is aggregate and store about 3% of India's total grain production, which is significantly large when you look at it on a standalone basis, but it is still only 3%. And I think one of the things that we realized fairly early on in our journey is that if we have to build an agriculture business, it has to be stationed and embedded at the farm gate and not away from it. So about 85% of all our business is at the farm gate in the primary and secondary agriculture locations. And there, what we do is primarily to see one bag of grain of one individual farmer. And how do we use that one bag of grain to offer an integrated set of services? So we first help them store the produce after harvest, then use that same crop, which is stored to give them a loan. And which is how by enabling storage and embedding finance, we help the farmer to actually hold on to the produce after harvest and not sell it immediately. And then the same produce is then published to a large variety of buyers when the prices are right and the farmer wants to sell and connect them and making sure that the farmer receives the payment in advance without having to go through the vagaries of market challenges. So I think the secret sauce, which is no, not a secret, but the real kind of element which has helped us build this business and achieve the scale is to primarily be near the farm gate, work with the producers and offer a set of integrated services, which ensure that the cost for the farmer goes down, whereas the income for them goes up.

Dhruv Sharma: Yes. So Prasanna, I think you cleared the question. I was trying to get to, which is it's mostly greens, maybe pulses. It's not perishables right now. The focus. That is right. Can you also help us understand when you say smallholders, how much land do they hold? Is it sub one acre? What's the definition?

Prasanna Rao - Co-founder, Arya.ag: So when we say smallholder farmers, we categorize them into primarily two sets of farmers. One is the small and the medium farmers. So the small and medium farmers basically hold less than five acres of land.

Dhruv Sharma: And so you must have had to integrate with farming communities to be at the farm gates. That is right.

Prasanna Rao - Co-founder, Arya.ag: So we work through farmer producer organizations, which are companies of smallholder farmers.

Dhruv Sharma: And how many crop cycles in a year are they able to, you know, go through with the smallholders?

Prasanna Rao - Co-founder, Arya.ag: So depending on which geographies they are in and what crops they normally grow, in India, we have three different harvesting seasons. So depending on the crop, say, for example, if it's wheat, it's a crop which is harvested once a year. So this is one crop in a year. If there are crops like corn, which is maize, which is harvested twice a year, then you would have two crops being taken in one particular season. Now, depending on the wherewithal and the resources that a particular farmer or a farmer community has, they take up multiple crops through the year.

Utsav Somani: And paint a picture of what the scale is like, the number of farmers that you have, what's the revenue of the business, number of warehouses, and through your NBFC, how many disbursements are you making?

Prasanna Rao - Co-founder, Arya.ag: So as I said, we, in a year, aggregate about $3 billion worth of grains, which is close to about 26,000 crores worth of grains are stored in our warehouses throughout the year, in one year. We don't take inventory risk. It is primarily, as I said, the crop that is stored by the depositors. The depositor is the owner of the produce. And the depositor makes a choice as to when and whom to sell that produce to. And of this $3 billion worth of grains that are stored, we give loans of about $1.4 billion every year, close to about 13,000 to 14,000 crores worth of loans are dispersed annually on the platform. And we help sell close to about 6,000 to 6,500 crores worth of commodities annually on the platform. As you sort of asked, we don't, you know, that's a philosophy that we've been very, very close to, saying that we are not another trader who would purchase from the farmer at a cheaper price and then sell it at a higher price, thereby, we making the money. Our thought is that it is the farmer's produce. We are happy being a platform. We're happy being a service provider, which ensures that all of these transactions happen on the produce, which is stored by the farmers, farmer organizations, small aggregators, and small traders.

Utsav Somani: So it's more like a managed marketplace. But what does an unhappy customer look like for you? Have you had those instances?

Prasanna Rao - Co-founder, Arya.ag: Of course, you know, and thankfully for us, we've not had too many of those. So an unhappy customer is that my promise or my, you know, pitch to a farmer is that when you harvest your crop, your, the prices at that time, because of the surge in supply, depresses down. And then to a large extent is almost sold at distressed prices immediately after harvest. And given the nature of the crop, you know, as the supplies dry out, the prices start moving up. So there is a delta in prices. Now, an unhappy customer is one who accesses my services and still does not make that incremental income as compared to what he would, if he or she would have sold immediately after harvest. So for us, we run a seven month report card business every year. That in seven months, the farmer, and we know if the farmer has made money net of the costs that he or she has paid to us. So if a farmer hasn't made that incremental income, then the farmer would never be happy and would never come back to us.

Utsav Somani: And what is the alternative to them? Like it's typically to go to a mandi and what cost do you charge them for your service?

Prasanna Rao - Co-founder, Arya.ag: So I think, you know, without Arya and with Arya is how I would sort of, you know, draw up the journey of a farmer or an FPO. Without Arya, the FPO or the farmer would have been forced to sell the produce at the time of harvest, at the prices, which are almost very, very low, almost distress kind of prices. With a practice where more often than not, they are exploited by the buyer because the buyer would not give them the right rate, would have, you know, other kinds of practices, which mar the earnings of the farmer. And the third basic thing has been that without Arya, close to about 8% of the farmer's produce is lost as food loss, which is generic data. And with Arya, basically the farmer has the choice now to store the produce for the money that the farmer needs, because of which he's actually forced to sell the produce. He gets a loan within five minutes of storage and can sell the produce when the prices are better after say about four to seven months and can choose whom he or she wants to sell that produce to from among various buyers who are bidding for his or her produce. Second, because of Arya, this food loss, which otherwise stands at about 8% is down to less than 0.5%. So basically, what we've seen over the last about 10 years of our work is that farmers on an average in a year earn anywhere between a 15 to 40% more than what they otherwise put.

Utsav Somani: Well, that's a significant enough jump. And I mean, in terms of geographies that you're servicing, is that something that you focus on certain regions or you're pretty agnostic?

Prasanna Rao - Co-founder, Arya.ag: So we cover about 60% of India's all districts. We're present in about 22 states across the country. Our presence in the northeast is lower and hence those are the few states where we are not currently present.

Utsav Somani: Anand, we've not given you a chance to speak. No, Arya, it's all good. Yeah, feel free to chime in. It's a very friendly chat. So whenever you want to say something or top up to what Prasanna is saying, I mean, the floor is yours totally.

Dhruv Sharma: No, no, absolutely. We'll do that. So because you gentlemen work in 22 states, help us understand, are there any, you know, are there any food supply related regulations? For instance, if a certain state is running a surplus, can it export grains domestically? How does that work internationally? Are we now like as a country, you know, running a surplus on certain commodities so that we can start exporting internationally? How does that work? Also, what's the FCA's role in all of this?

Anand Chandra - Co-founder, Arya.ag: Yeah, I will split this into two parts. One is the role of government. So government basically focuses on buying and converting that into rice. No, I can hear you. Hello, am I audible?

Utsav Somani: Okay, I think the internet might be slightly glitchy. Yeah.

Anand Chandra - Co-founder, Arya.ag: So paddy is and wheat are the two major commodities which FCA buys basically from the food security per se and for the distribution under PDS, the public distribution scheme to the ration holders. They procure almost 25 to 30% of the total production of paddy and wheat and then distribute it. If you look at the entire agroclimatic zones in the country, each agroclimatic zones helps us to produce some commodity or the other. Now domestically, say for example, wheat is produced mostly in the central and the northern parts of the country and for the requirement of maida or atta, it gets shipped from the north to the south. So that's how it happens. Similarly, maize gets produced mostly in Karnataka, Andhra but gets consumed mostly in Tamil Nadu because Tamil Nadu has helped set up a lot poultry units for cattle. For poultry feeds. So domestically, most of the commodities move in and around based on the consumption pattern. For example, tuvar gets produced in Karnataka and Maharashtra but is consumed as sambar across the country or to arad dal, what you call it. So domestically, it gets done. When it comes to export, basmati rice is one of the largest commodity which we export, especially to the Middle East.

Dhruv Sharma: I'm told it's sold in Texas as taxmati. Some version of basmati rice.

Anand Chandra - Co-founder, Arya.ag: See our export to US is very limited on the basmati in European countries because of the stringent norms they have. So basmati being the largest, most of the other commodities, we are not so competitive when it comes to pricing. So we find it difficult to match the corn or soya prices from US, Brazil, China. Cotton is another agri commodity which we export. So it depends on the demand and supply. We are surplus in basmati right now. But the India-US tag for is slightly linked to what is happening on the agri sector because once you open the floodgates for corn and soya, the amount of dumping of those commodities which will happen in India will make Indian farmers suffer more because the prices will go down and then anyhow we are small and marginal farmers, it will impact the income more.

Utsav Somani: Two questions. How much of your business is import versus export? And also, are there any certain grains or commodities that you do not touch? You mentioned some of the big ones, but are there any that you stay away from just because of the volatility or other factors?

Anand Chandra - Co-founder, Arya.ag: As Prasanna mentioned, we don't get into buying and selling commodities, but we do facilitate, say, basmati rice players. We help them store commodities and facilitate funding to the farmers and some bit of the small processors who in turn export those commodities further by processing it. Similarly, pulses is one commodity which India imports, especially the yellow peas and the lentils because India is deficient in pulses. So we import. So we help the importers at times at the port to store the commodities and then facilitate the entire trade. We directly are not into import and export. Some of the major companies remain paddy, wheat, maize, pulses and oilseeds. Though we have a list of another 30-40 odd commodities, but they are smaller in the city.

Utsav Somani: So with the global large grain traders like Cargill and ADM, like do you interface with them or that's not something as a focus for you?

Anand Chandra - Co-founder, Arya.ag: So while we started, we adopted a cluster-based model. So our endeavor was to layer services on. So we started with storage as a services. We added the layer of finance on it and then we added the layer of paid on that, the commerce piece on it. So why initially, just to take a minute on the background, when we started, we had a capital of around a crore only. This is 2013-14. And a major of our competitors who are in capital base with them would be 100 plus crores. So we went to the smaller towns, smaller markets. So while we went to the smaller market, we gave a proposition to the corporates like Cargill, LD, Glencoe, ITC, that your cost can also be minimized if you buy from smaller mandis. And it also gives an opportunities for farmers to access players like you. So we gave them a warehousing proposition in those territories, created a market for us by working with those large corporates. We spread our names and then we went granular in those markets to set up our finance and the commerce franchises in those setups. So we interact with them. 15-20% of our total storage would be held by these corporates with us where they take only storage services, but they help us in buying commodities from the farmers through the platform. So they may not take finance from us, but they take storage and the commerce bit from us.

Utsav Somani: Interesting.

Dhruv Sharma: Yeah, I had some questions on the financial side. So you mentioned loans. Specific to loans, are there end-use restrictions? Can farmers only use those loans to buy inputs for the next crop cycle or how does that work? And I'm also curious, what are the other opportunities in terms of financial products? I know crop insurance tends to be heavily politicized at times, but for private players, what are other opportunities in terms of bringing financial products and services to the agri space?

Prasanna Rao - Co-founder, Arya.ag: So in terms of the farmers accessing these loans, the end-use is primarily their consumption requirements. So if you see when you offer post-harvest credit, the part of that credit goes for extinguishing the pre-harvest credit the farmer may have taken, maybe locally from local moneylenders, maybe from other banks, et cetera. A part of it also goes in for their consumption requirements. So that's how the whole requirement with respect to their end-use. But as a lender, what we basically see is that since this is a loan which is given against the crop that the farmer stores, the end-use is mostly limited to requirements or funding needs that went into the production of that particular grain. So that's how that has been. With respect to the other opportunities that we have seen, one is credit, both pre-harvest and post-harvest is a very large area. The other bit is that when you work with farmer organizations, it's there. So the working capital requirements of the farmer organization, they offer inputs, say seeds, fertilizers and other kinds of inputs to their members. So there's a working capital requirement there. So specifically financial offerings which cater to the working capital requirement of farmer producer organizations is the other. Similarly, there are other kinds of services which cater to insurance needs. You mentioned about, say, insurance being at some points or sometimes being a little abused as a word with respect to farmers, but it is still a very, very important tool to mitigate risk at the farmer's end. So as long as you're catering to the genuine needs of the farmers and embedding that in the farmer's life cycle, it is a very, very valuable offering. So we've seen each of these aspects playing out. One of the things that we have seen at our end, and I think one thing that Arya has pioneered in this entire space has been the ability to ensure price risk for farmers. Like one of the questions when you asked, what is an unhappy farmer? The unhappy farmer is that if the prices were to increase and they didn't, and the prices actually fell, then who took that price risk? And which is where we've been able to work with insurance companies to offer price risk insurance to farmers and FBOs. So that if the prices go down, there is a protection that the farmer has. So there are many of these innovative and simple kind of solutions that really offer a very large potential in the segment that we work in.

Utsav Somani: So Dhruv and I were joking in the world of AI, either you're building like hard technical skills or you're living in a mountain somewhere. So, I mean, your business is definitely sort of this kind of business that is resistant to the world of AI, right? But I'm guessing you're using technology in some of this stuff like the price prediction and demand prediction and other things as well. Is that something that you focus on?

Prasanna Rao - Co-founder, Arya.ag: So technology for us has solved for two basic problems. One is access and second is assurance. So when we say a loan is disbursed in less than five minutes of storage, that cannot happen without technology. So all the loans that we disburse on the platform on our own balance sheet are disbursed digitally and to the kind of customers that we are talking about. So there's a great deal of adoption that we have been able to drive with respect to technology on the ground. With respect to AI and...

Utsav Somani: How does that happen? Actually, that might be a very fascinating topic in itself that we can spend hours on, but tell us two, three things that you've done on ground which you're proud of in terms of activating and educating and...

Prasanna Rao - Co-founder, Arya.ag: So the first thing that we've done is to digitize agriculture warehouses in the country. So while there are about say 120 to 130,000 agriculture warehouses spread across the country, we have about 12,500 of them digitized and discoverable on aria.ag. So it's like an Airbnb of agri warehouses, if I may well. So that's the first thing. So you could go out, identify a warehouse, book a warehouse online, where each of these warehouses are sort of digitized in a manner where each warehouse is described on more than 100 parameters, which a user can actually come and identify and book. The second thing is that when any grain comes into the warehouse, the quality of the grain is assessed using AI and image analytics. So it's a double-sided scanner, which is pending a patent from our side, where just by putting in the grain, immediately, instantly, you could have a quality certificate generated from it, which cannot be disputed later. Now, on the basis of that quality certificate, we value the commodity. And all of that is instantaneous, and which is why once you value the commodity, the loan is say 75% of the value of the produce as on that date. So the moment the produce comes into the warehouse, the quality is assessed. You can actually disperse the loan into the farmer's account in less than five minutes, because of these interventions.

Dhruv Sharma: Prasanna, may I ask a quality-related question? Because maybe now's the best time to ask it. So when a really large buyer like IT wants to buy a huge supply of grains, like all biscuits must taste the same, though not all farms create the same kind of output. How does that work?

Prasanna Rao - Co-founder, Arya.ag: So grains, fortunately for us, while it appears to be complicated, is not as complicated as it appears to. Grains are graded into different categories. So purchasers come in for specific grades of commodities. So an ITC would need a certain grade of wheat. Say a poultry manufacturer would look at a certain grade of maize. Suppose there is an animal feed manufacturer, they would look at a specific category of wheat. And which is how, what really happens at the warehouse before the commodity comes in, is to assess the quality and grade them into different categories. And hence, the power of a platform is to actually match a specific grade to a specific buyer. Now, if somebody wants human consumption sweet corn, now that is something that may not be available across the country in all warehouses. So you need to be completely aware of these different grades and then find buyers for each of those respective grades. And the power is, how efficiently can you reach out to those category of buyers? So that is where I think ARIA has differentiated itself, to be able to first identify and work with a wide set of producers, which are farmers and farmer organizations, then grade their produce into different categories, and then find buyers for each of those categories. And which is how your offering becomes complete.

Utsav Somani: Prasanna, there's one final question before we let you go. I think our next guest is waiting. I mean, fascinating what you've built as a business and kudos to you and the team for getting it to this stage. There's a question, I'll paraphrase it a little bit. Why have agri-tech startups not been able to make the cut in India? Is it the general nature of the industry which likes it to be traditional? Then they mentioned a name which has fizzled out in the last two years.

Prasanna Rao - Co-founder, Arya.ag: Anand, would you want to take this?

Anand Chandra - Co-founder, Arya.ag: Yeah, see, some of the agri-techs have not been able to do a grounds up on the requirement of the market, basically. People have adopted tech which have worked in the Western countries or the European countries. Say, for example, some of the tech companies have started on two major fronts, where one was demand aggregation and second one was on data-driven approaches. Trust me, on the second front in India, we still are way behind where people will start paying you for data. Second, on the first point, which was more of the ground-up approach, people have used tech to generate demand, but the real challenge lies in servicing the demand. Now, you look at it from the perspective, when we talk about 85% to 90% of our farmers being small and marginal, a farmer may need one bag of media which may cost 1000 rupees. To export or to courier 1000 rupees bag will cost you 300, 400 rupees. Where will the margin come? So, the entire agri-techs have focused on trying to corporatize the entire supply chain. What we believe in that these supply chains are, or the middlemen which we call as villains are actually not villains. They add a lot of value in terms of aggregating and even distributing. Aggregating the output and distributing the inputs. Corporatizing this has not worked out because the cost goes up drastically. So, what this industry needs is a transparency and availability of credit at the right moment. So, once you solve those two bigger problems, it will fall in place. Trying to replace from say farm to folk or trying to replace those input, trust me, these input companies like Poor Mandal, Chambal and all have been working for 20 years now and they also have tech backings. The reason why they've not been able to move to more tech enabled because of the requirement which is there for the farmer, which is small and marginal farmers. Unlike the European countries where one requirement from a farmer would be one truckload of urea, which becomes easier to monitor and dispatch.

Utsav Somani: All right. I think that's a good note to close on. Thank you so much for joining us on this show. Thank you.

Dhruv Sharma: Real pleasure. Thank you.

Utsav Somani: Thank you. All right, folks. Now, we're welcoming our next guest, Prateek. Prateek, welcome to the show. Hey, Prateek. I think you're on mute.

Speaker 6: Yeah, yeah, yeah.

Utsav Somani: Here's the middle. Thank you. The thing that we all run into when it's most not needed. For our listeners, we're hearing about Bholna for the first time. I love the name, by the way, truly Indian. What does it do?

Prateek Sachan - Founder, Bolna: Yeah, so we enable companies and people to create and deploy their own voice agents at scale. Okay. So any company who wants to leverage voice capabilities in their own workflows or applications in any way, they can come onto our platform and use our APIs. Simple as that. Instead of doing the entire thing on their own end. So we handle the entire pipelining of voice agents, deployment of voice agents, scaling of voice agents, everything. And this works across industries and sectors? Yeah, yeah. So we are acting as a horizontal platform. Any industry, any use case, any vertical, they can and are able to create agents on top of us.

Utsav Somani: And congrats on the raise from General Catalyst. Yeah, thank you. So, I mean, for our listeners, we're hearing about VoiceAI, like I think everywhere. Like VoiceAI is extremely important. Can you tell us the landscape? Like what kind of companies exist? Like a more macro picture on what are different companies doing? And why did you land on Bolna as a company?

Prateek Sachan - Founder, Bolna: Yeah, so true. I mean, VoiceAI is currently one of the hottest thing in the market, right? I mean, every company is trying to build something in voice right now. Or maybe pivoting to voice, something in voice, right? So there are now multiple layers, right? So one layer is the bottom layer, which is the foundational models, which the OpenAI, 11 Labs, all of these, you know, Frontier Labs are developing their models, right? On top of that, I mean, on top of the foundational models, are players like us, which are building a horizontal platform, where we are basically mixing and matching different models, depending on, you know, the capabilities which each model gives, right? Because OpenAI might be good in one, like, you know, some things, but it might not be good in other aspects, right? Similarly, it's the thing with other models as well. On top of us, there are now companies which are building, using our APIs, niche vertical. So let's say someone is building voice agents for FinTech, where they are doing, you know, collections use case. People are using us for recruitment use cases, right? Like where you can get a call or a candidate can get a call, who might have uploaded the CVs or resumes, right? So now there are niches and verticals on top of us as well. So these are primarily the three layers. And of course, I mean, there are the chip companies as well, right? Which are sitting behind the foundational models companies.

Utsav Somani: And in what timeframe do you see that I'm talking to an AI bot? I'm probably already talking to an AI bot, if I'm calling HDFC or anyone, but when do you think like we replace 100% of that layer?

Prateek Sachan - Founder, Bolna: I don't think it is a question of replacing the 100% layer to be honest. It will only be, you know, how to focus on few aspects versus other aspects, right? So for example, you took an example of HDFC, right? So now, for example, HDFC without AI or without the voice AI into the thing, their teams might be calling a lot of people, right? A lot of people might not be interested in whatever HDFC is selling. Maybe it is cards, loans, insurance, etc, right? So now with the use of voice, they can actually automate few standard calls and then focus on those 5% of the cases where a human is actually required to call the person, close the deal, whatever is there, right? So a lot of use cases are getting unlocked in that manner.

Dhruv Sharma: Prateek, can you help us understand how maybe or rather simplify for us technically how you get conversational agents to develop like contextual awareness? If I feel grieved as a customer, I'm naturally upset and if I'm talking to a bot who has a very frivolous tone, I'm not going to like that in a moment, right? So maybe with that as an example. Emotions, yeah. Emotions also. Yes, I mean, we have so many questions for you, but let's just do it one at a time, yes.

Prateek Sachan - Founder, Bolna: Yeah, so that is where we actually come into the page because every use case is different, right? So for example, when you say, you know, a very unhappy customer who has paid something but the payment has failed, right? You need to be patient with that person, right? Compared to maybe a HDFC kind of a use case where it's more like sales kind of a thing, right? So we give functionalities where, you know, people can tune their agents where they can choose, you know, like how responsive or how patient you want the agent to be. So whenever you are talking to a customer care, you know, like type of a use case agent, the agent will be very, you know, like calm, it won't be interrupting you in between, it will listen to you, you know, it will give you more time to, you know, to say things out loud. So all of these things we have actually enabled people to use.

Dhruv Sharma: Is that, Prateek, is that what people mean by orchestration? I see that term being thrown left, right and center. Does it mean model selection? Does it mean finally, you as a practitioner define it in very, very simple and vivid terms?

Prateek Sachan - Founder, Bolna: Yeah, sure. So I mean, so there are different kind of models, right? So one model is, let's say, just in charge of, you know, converting my, you know, whatever I'm speaking in textual format, right? There's another model, which is, which, which handles the, you know, the reverse of this thing, right? Converting the text into audio, right? Now, how do you combine these two things? So that they work very closely together, right? So as soon as I'm speaking, the textual format is going to another layer, where, you know, a lot of, you can say a black box is there, where it is capturing that, okay, I'm unhappy because my payment has failed. And now I'm worried whether, you know, that payment will get refunded or not, right? So automatically, since I'm a customer of that company, I will have some order ID. My phone number is already, you know, in their system. They will actually check using my phone number that, hey, the order ID, which is mapped to this phone number, you know, some API calls will be there. Then, okay, this order got failed, right? And the refund will be processed in the next 24 hours, for example, right? So this is the textual content that I have got, right? Like, like, like my system has got. Now it will again get converted to, you know, in a good soothing voice, where the agent will actually tell that to the customer itself that, hey, we got your payment, but your payment was failed. Please wait for 24 hours and you will hear back from us or you will get the amount refunded automatically.

Utsav Somani: And tell us a little, I mean, if I'm a business, I'm coming to you versus say, suppose I have a call center of 10 people making these calls versus a business coming to you. What is the cost differential between these two upon just the efficiency boost? What is the cost to a business for using?

Prateek Sachan - Founder, Bolna: Yeah, so we have seen like using voice agents, like for us. So we are pricing it around 4 to 8 cents, depending on the volume, use cases, the choice of models. And we have also seen that if you go to a fully fledged call center per minute, per minute of usage, yeah. And then you go to a standard, the existing platforms, the existing solutions that are there, you know, like from hiring to the office space or to the onboarding of a team, company, all of those things will actually make the price to 12 to 15 to 16 cents as well, right? And then you will have to also manage, you know, people leaving the company, new people are joining, you are again onboarding them. So all of those things are very operational, right? Which has some additional costs attached to a particular company or a particular, you know, segment.

Utsav Somani: I'm gonna come to a very India specific problem, which I think is becoming very common in India that, I mean, if you can spoof somebody's voice, it is a chance that you can convince somebody who's not educated enough or is not aware enough in that moment to actually give out important information like your UPF or OTPs or any other stuff. Like, I mean, isn't the world gonna be scary when all of this starts happening? It's true.

Prateek Sachan - Founder, Bolna: I mean, yeah. I mean, this is one of the things where we are also working on how to, you know, evaluate that the people who are creating voice agents are actually for authentic use cases, right? They're not using any, you know, things to do any fraud, no political campaigns, for example, no old calling, for example. All of these things are actually taken care of through certain checks and compliance as well as through tech-driven systems. So both the things we are making use of.

Dhruv Sharma: Is it possible to discern between synthetic and authentic voice, Pratik, with like a reasonable degree of confidence?

Prateek Sachan - Founder, Bolna: So, I mean, for us, what we have seen is, so whenever we are calling people and those users who are actually receiving the call, right? They are not in tier one cities, right? They are actually not able to make out whether it's AI or actual human beings, you know, spoken to. People that, you know, that is being called, you know, India one, you can say India one segment. Those, of course, they can know that, okay, this is some AI being called, right? Because they are in this space, they know this domain, right? So they can actually make out that, okay, this is an AI or this is not an AI.

Utsav Somani: No, Pratik, what I was getting at is if your machine itself can tell, like if a bot is making call to another bot, will the bot know that it's an AI call?

Prateek Sachan - Founder, Bolna: Yeah, yeah, so yeah, yeah, yeah.

Utsav Somani: So actually, yeah. Agent to agent, basically. Agent recognize agent or no?

Prateek Sachan - Founder, Bolna: Yes, yes, so it actually does. So we are doing certain use cases where, you know, people have created multiple, you know, bots, where an inbound call is actually calling, is actually being called by an outbound call, outbound agent, right? So, for example, these, you know, on-call support that have, right? Where a system goes down and for one minute, if there's no developer, then a phone call will go, right? So now over there, people have brought in a voice agent in place that first the voice agent will actually check whether the system is actually down or not and only then call the actual, you know, developer who might be asleep at three or four in the morning.

Utsav Somani: Yeah, even the latest iPhones have that feature, right? You can screen before you answer. So you have 800 paying customers. What are the big use cases? You've got some big names like GoQuick, Align and like many other large organizations and startups. So what are the big uses?

Prateek Sachan - Founder, Bolna: So it's actually scattered, but the majority pushes, you know, into e-commerce. Then BFSI collections, you know, this fintech segment, you can say. There are a lot of inbound use cases where it's majorly customer support into multiple verticals, right?

Utsav Somani: And do you train on that actual data as well? Like, I mean, if suppose you're doing it, a customer support post e-commerce sale for GoQuick, will they give you access to that password? Because whenever you hear that, when you're making a customer support call, you hear that this call might be recorded for our quality and training purposes.

Prateek Sachan - Founder, Bolna: Yeah, so we act as a platform. So GoQuick has treated the agents on using our APIs and they have given, like if they want, they can give access to their APIs. They can tune their agents in that manner where, you know, in the real time, they can query GoQuick APIs to gather some information if there's a need to gather them. So we just act as a platform. We provide the functionality, the capabilities, and then companies and people use our APIs as a platform to build solutions on top of us.

Dhruv Sharma: Prateek, I'm curious, how are, you know, owners of IP thinking about voice? Like in general, like, you know, publishers and gaming companies and so on and so forth, the music industry.

Utsav Somani: I mean, yeah, so there are... I mean, in terms of IP, you can click a picture and actually create a whole world now, a table. I saw that video where a cat, somebody took a picture of a cat and you can now control the cat in a loop.

Prateek Sachan - Founder, Bolna: Yeah, I don't know, what's it called? That I'm not sure, but even, you know, the recent thing with Twitter, for example, right? People were posting photos and just, you know, doing random stuff around the photos. In case of voices also, right? You can upload, you know, snippets of voices, snippets of, you know, like specific version of voices and you can clone a voice, right? That is it. But yeah, I mean, still it's a very evolving field and I'm pretty sure that, you know, there will be, you know, platforms that will emerge who will help us, you know, to predict that thing, to guardrail that thing as well. So for example, we make use of 11labs voices, for example, right? And 11labs has partnered with certain actors to get the voices cloned as well.

Dhruv Sharma: There was one actress who got really offended with the opening.

Prateek Sachan - Founder, Bolna: Yeah, very, very, very early back.

Utsav Somani: Yeah. And I think it's been an issue in the world of Hollywood as well, but I think all the actors protested and said that, I mean, the likeness of like somebody living on via AI, I think it's going to be an issue, right? I mean, if you're, I think personal copyright will, or your rights will become, I think, even more important. Yeah, they will be challenged for sure.

Prateek Sachan - Founder, Bolna: Even in the beginning, there was some case where Anil Kapoor, like, you know, something related to Anil Kapoor was copied or was not used in a specific segment. I'm pretty sure. I like, I read some news around that thing.

Utsav Somani: All right. So as a final question, we read about your Y Combinator journey. I think you tried to get into them, the accelerator for five times. Is that true?

Prateek Sachan - Founder, Bolna: Yeah, around four to five times.

Utsav Somani: Yeah, yeah. How did you manage to convince them that Indian businesses are ready to pay for your service and business?

Prateek Sachan - Founder, Bolna: Also, that's a funny story. So the initial time when we applied to Y Combinator, it was exactly the same idea that we are doing right now. But between the first time and the, you know, fifth or sixth time, we made small, you know, small soft pivots in the voice AI domain. So we, so once we did, you know, voice AI only for Shopify brands, then voice AI for frontline workers, then voice AI for hiring, then voice AI for sales. So these are three, four segments that we tried out pretty quickly. But every time we were using our own platform, right? So in this timeframe, our platform also improved a lot. And 2025 was when we saw, you know, that people started using us a lot in a very self-serve manner. They were actually paying us. And we had good revenue to, you know, to prove it to people who were just saying, you know, that, okay, India won't pay. No one in India is going to pay so much of, you know, money, people don't understand AI from India. All of these things we used to hear, right? In our journey. But then we showed them that, okay, there's actual revenue being, you know, being made. There are actually people, you know, two, three person company who are actually doing so much of volumes using us from a small company point of view, as well as big companies that are using us, right? So we have a curve covering the entire spectrum of companies from small companies to middle companies to public sector companies, like all of companies. So these were very strong, you know, signals which actually played, I guess, in our case.

Utsav Somani: Kudos to you, Prateek, for all the endurance and wishing you all the best and congrats on Money in the Bank. Cheers. Thank you.

Prateek Sachan - Founder, Bolna: Thank you, sir. Thank you. Thank you all.

Utsav Somani: Thank you so much. All right, listeners, that's it. Let's head into the weekend. We'll see you when it turns February on Monday.