InterviewsNews & AnalysisNewsletter

Built the gold standard when it comes to AI on local language content across formats: Umang Bedi, Dailyhunt

umang bedi

Tech veteran Umang Bedi hardly needs any introduction in India. Bedi, who is co-founder of the largest local language content platform in the country Dailyhunt, has held several senior leadership positions in his illustrious career. These include those as MD of Facebook, India and South Asia, MD of Adobe, South Asia and MD of at and emerging markets at Intuit, Inc. Dailyhunt, as we know is part of Bengaluru-based Verse Innovation, which completed a $100 million+ round of funding in December 2020 from AlphaWave (a unit of Falcon Edge Capital), Google and Microsoft and its existing investors–Sofina Group and Lupa Systems.

The company, then, raised another $100 million in a Series H round that was led by sovereign wealth fund Qatar Investment Authority (QIA) and Glade Brook Capital Partners, with participation from Canaan Valley Capital and existing investor, Sofina Group. This has raised the company’s value to over $1 billion, making it a much-envied unicorn.

In an interview, Bedi spoke to Leslie D’Monte–Publishing Director of Trivone, about many issues, including his current journey as an entrepreneur; the role of tech startups; Dailyhunt’s business model and prospects; the role of AI in this business; and the role of the tech ecosystem in India. Here are edited excerpts:

* You have been at the helm of many billion dollar global companies. Now you’re heading an India-based unicorn where you joined as a professional but now are an entrepreneur? Has this changed your approach to business and strategy?

The global multinationals that I’ve worked with are fantastic businesses and great companies, and each of them is very well regarded. But the differences that I see in those roles are: First, you have a very large safety net within your corporate brand. Two, you have very little control over the product because the products are largely created for global audiences out of headquarters. And three, the India businesses for most multinationals are usually go-to-market engines. So it does give you a lot of promise to go and build businesses that are focused on India, but they’re largely go to market operations.

The journey as an entrepreneur, on the other hand, is very, very different. Firstly, I think the biggest difference for me is to have a co-brother Viru (Virendra Gupta), who’s my brother in arms. And to have a partner whom you can trust; have the deepest professional and personal relationship with; and complete intellectual honesty, is a very, very big difference from the world that I come from. I think the second key business about entrepreneurship is about doing more with less–you’re always going to have limited resources, you’re always going to have limited capital, and limited of resources in terms of marketing, people, and costs too.

So you’ve got to start delivering a lot more impact with less, which allows you or forces you to ruthlessly prioritise on the core things that matter. So I would say that’s a very big key second difference. The third difference which is truly liberating, is that you control your own destiny–from the point of view of a product and offering, you control the end-to-end roadmap of how your offerings could look like. You control the end-to-end strategy of how you want to grow the company; you control the entire destiny around how you want to raise capital and create stakeholder value for your investors, shareholders, employees, customers and business partners. So it’s far more liberating and a lot more fun.

I think the fourth difference, which is very humbling is that at the end of the month, you always have to have enough capital as a runway, right. And so you have that responsibility of a large team of ensuring they’re kept happy. And I think that responsibility lies on my shoulders too. And I saw, and I think that that’s very humbling in its own way. Because valuations at any point of time are just a reflection of a particular value of a business. But there’s a human factor of people, and you feel far more directly responsible without that safety net.

Hence, this is the best decision of my life (becoming an entrepreneur). I regret doing it at 40. I wish I’d done it 10 years earlier. Yeah, but now you’re getting to shape your own destiny, as you say, Yes, absolutely. I’m loving that.

When you’re with a multinational, you have shareholder pressure. As an entrepreneur, you do have investor pressure. That said, most founders and investors of tech startups, especially unicorns, appear to be playing the valuation game–they seem to be chasing revenue and adding consumers by burning cash and accumulating losses without working towards making profits or even breaking even. Would you agree with this criticism?

My reaction to that, Leslie, is that there are two kinds of businesses in India. And there are two kinds of founders. And there are two types of investors. And there are two types of directions that I see in the market.

In the specific list of unicorns that you talked about, and there are over, you know, close to three dozen unicorns, two or three dozen unicorns today. There are a lot of businesses, which are really, really solid, and are generating cash, and getting to break even, if not breaking even. These are businesses in search of a valuation–they could be a lot larger in terms of value. But I also see a lot of businesses that are pure valuations in search of a business or a business model. I believe more in the former.

I believe there are two kinds of businesses–if you are a business that makes a great product that is growing virally, people come back and spend hours every day or use the product organically. It’s all being driven by word of mouth–you have virtually zero to no marketing expense. This is just right–the user growth and engagement growth journey. And I don’t mind such businesses not generating any revenue because they’re just such awesome product experiences. A great example is what you and I live on on–WhatsApp. It’s a great business. It ultimately was bought by Facebook for you know, close to $20 billion dollars. It is a viral app–they spend zero on marketing. It’s a small team. It has never generated revenue.

However, what I see most of the time, are companies that are creating products that are not viral; they’re burning a whole lot of capital to try and get people to come and download and adopt. They have a very leaky bucket because of which retention is very, very low. And they’re not generating any revenue or any cash or any such metric. I worry about these businesses because it’s perfectly okay to not be revenue generating–I think revenue can always come later in a in a digital business. But then the core unit economics of customer acquisition of lifetime value of retention is something we need to look at really. I am more pessimistic about the businesses and their long runs because they’re riding a wave of investor sentiment. And right now, as we know, capital markets are flushed–the US is printing money. And there is a lot of liquidity.

I’m a big fan of businesses that are real–that are on a path to generating revenue or a path to profitability, and are showing the right unit economics even if they’re burning some capital and spending money on marketing. That’s fine. But these are more sustainable.

In this context, what’s your business model and strategy?

Our business model is actually among the simplest on the planet. Versa Innovation, as I said, is a tech platform company that’s in the world of producing algorithms. We produce algorithms and power. Dailyhunt–India’s largest local language content app–has licenced relationships with all publishers for its content. And it has a large user base that spends time, which creates advertising inventory. We sell our inventory, and make revenue, which is passed back to the publisher. So it’s a very fulfilling flywheel. with Josh, we’re on that journey as well. We have a large number of content creators on the platform, across large, top of the line influencers–and they’re creating content. And as we start monetizing, and building more revenue streams across it, we keep sharing revenue with the creators. So it’s a very self-fulfilling business model. The company is already generating over $140-150 billion of ARR (annual recurring revenue), and it is growing by more than 50%. So overall, we have a very solid business with a core business model.

How is your strategy shaping up–you’re growing organically, and you’re making acquisitions?

Our strategy is very simple. We want to own the mindshare, timeshare and revenue share of all local language audiences across a family of apps. In the next four years, when the Internet has a billion people, we want at least close to a billion to be using one of the couple of apps that we will launch. We already have two, and there’ll be a couple more we will be introducing in the future. We expect at least 500 million people to be coming back every day and using these apps, which we will continue to monetize.

We make acquisitions in areas that enhance the value of our business. For instance, Vebbler comprises a very small but very young and talented team that came on board. Sahil, who was its founder, now leads the team at Josh. He is a very young millennial product head, who understands the audience and is a creator himself. Cognirel, too, is led by an exceptional leader in Ramprakash who has a very deep AI-led understanding of local language content. He’s also a linguist, and he has turbocharged the journey of building our AI Labs–that is going to provide the right set of algorithms to these family of apps to power our content personalization and content discovery journey. Hence, the acquisitions we’ve done till now are more around skill enhancements versus getting into new markets or getting into new products.

Kindly tell us more about your AI Labs…

I would like to believe that we are at the cutting edge of probably having built the gold standard when it comes to AI on local language content across formats. We today run a data set on AI, which is over 6-7 petabytes, and is growing by over 8-10 terabytes every day. Essentially, we have our learning models being trained on over 250 million pieces of content across audio, video, imagery, and text across 14 local languages. All the content resides on our platform. Then we have 80+ proprietary AI/ML tech models that run to understand the content and the context. So essentially, we understand whether it’s the content is in a certain language, we look for adjacencies of keywords to understand genre, we then understand micro genres.

For instance, if we are making a comparison of three mobile phone players, we look at the social media embeds in those posts. If there is a video, we open up the video, break it up by frames to understand the content–we isolate the audio and look for pattern recognition in terms of what it’s saying. We look for voice-to-speech conversions to really understand that language. We pick up keywords; we look at the content and check whether it’s safe for viewability. We do that with images as well. And all in all, when one piece of content is ingested, there is like a multi-tiered level of knowledge that is developed on that content across a taxonomy that is then generated with hundreds and hundreds and hundreds and thousands of tags describing that content.

On the other hand, when a user comes on to using the app, we don’t force the user to log in. He or she can just start consuming content. But the manner in which that anonymous ID consumes content–in terms of the way he or she scrolls and spends time, we get an idea of the click through rate, (and the person’s) likes and dislikes. We, then, have the ability to understand what type of content they like to consume–from whom, and at what time of day. The AI basically makes the decision of marrying supply with demand–the supply of content from those genres and types to the kind of implicit preferences that the user has that we are deriving from the user’s behaviour or any explicit signals that s/he has given us.

So AI plays a role really in three parts. One is understanding the content and the context. Two, is understanding the behaviour–both implicit and explicit. And three, the decisioning by which you marry the supply with the demand to create a very personalised, meaningful feed for each of those hundreds of millions of users who use the product.

Given these abilities, what’s your overall long-term vision for vernacular content in India?

The long-term vision is very simple. When you look around the world, barring China, there are no markets where large multi-billion dollar tech digital media businesses have been created. Because in most of our all of those markets, the duopoly of the global tech giants dominate–they dominate the time spent and they dominate the monetization. Our belief is that the internet is going to mirror traditional media consumption behaviours. Today in India, there are over a billion people who consume television media–almost 90% of that consumption is not in English. If I take Hindi aside, which is nearly the remaining half of around 40-50%, the rest is regional language consumption.

When we refer to regional languages, it’s Hindi and those languages combined. The internet has around 550 million active users today, about 250 million of who consume content in English, and a little more than 350 million in local languages. Nine of the 10 new users that come online every day are local language audiences. By the time the Internet has a billion users, about 750 million will be local language users. That’s the area that we want to serve. It doesn’t mean we won’t serve English, which is a content choice on our platform. But English is largely used as a dual language–I could be reading, or watching Malayalam videos and English or reading Telugu and English.

We’re very comfortable in our local languages–that’s the behaviour we want to reward. Our long-term vision has also spurred the local language creator community across organisations–professionally-generated content creators, individual content creators, influencers, and UGC (user-generated content). We want to fuel that ecosystem, and match it up with the demand and build a really large tech platform-based business that is a credible third front to the duopoly in our country.


But what about the smaller competitors in the space who also are growing at a faster at a fast clip?

The market is massive. And I’m a big fan of Indian tech entrepreneurs who are growing. I love the fact that so many young engineers want to build companies and not be in the safety net of a job. What they’re doing is very brave, so I have huge respect for the community. And I learn from the community every day–there’s so much reverse mentoring that we get every day. Real businesses are built on three fundamental rights to win, no matter what business you are in. One–that you have ability to attract a large audience. Second, have the ability to engage and retain that audience. And three, have the ability to monetize that audience–whether it’s through advertising, commerce, subscription models, etc.

What I’ve seen is there’s a large market, but it’s also a market that demonstrates a winner-takes-all characteristic. The top two or three players have 70-80% market share. I think the market is large enough for two or three players definitely–Time will tell who those will be.

We like to believe we will be among them. But I believe it’s going to require very fast execution of core modes on the business, real business models for those businesses to actually compete and succeed in the long run.

What more can be done to promote and encourage the tech entrepreneurship ecosystem in India?

The largest help we can do is to encourage entrepreneurship is by including it right into our education system. We don’t do that. I remember the diktat for me was to finish school, finish college, and get a job. I think entrepreneurship needs to be encouraged, right from the early stages. It’s a change in mindset, because you need to not be punitive towards failure–as an entrepreneur, you are going to fail. So you need to be able to celebrate failure, learn from it and encourage it. It’s important to fail, but fail fast.

The second thing is that we really need to start thinking about how we develop great product companies out of India similar to how we had that wave of building amazing IT services companies. Again, the framework for innovation needs to be built and encouraged at all levels–right from school to college, to innovation mechanisms across the industry. The third thing one needs is to have the ability to get access to Indian capital. It’s a bit of a shame that all the unicorns have been built out of foreign capital, whereas there’s so much Indian capital being deployed into private into public markets. There has to be an impetus for Indian funds to really start winning in this in the private market space as well.

Is it that Indian investors do not believe enough in Indian entrepreneurs?

No, I don’t think they’ve looked at this as an asset class yet.

Do you see India becoming an AI superpower? I’m asking this question is because I see this happening more in the AI application space simply because India does not have semiconductor plants or the R&D to compete with the likes of China or the US or the UK…

I definitely hope that we go down that direction, but I feel it’s more of a dream or a hope right now–a lot more needs to be done. We do have some of the best engineering talent in the world. But it’s not talent that has really delivered great outstanding products. It’s talent that has been used around delivering great services. I think, to even become an AI superpower on applications, we’ve got to start creating awesome product offerings. We need to nurture that whole ecosystem very differently, and give it access to you know, Indian capital.

Once again, thank you very much for your time. It’s always been a pleasure talking to you. I wish you all the best and hope your company becomes a Decacorn or even a hectocorn.

Leave a Response