In this episode of The Media Room, Vanita Kohli-Khandekar speaks with Toshit Panigrahi, CEO - Co-Founder of Tollbit, for a deep dive into how AI is reshaping the economics of digital publishing and content monetisation.

Toshit explains how large language models and AI platforms are changing the way content is accessed, distributed and consumed - often without directly compensating the publishers who create it. He breaks down the growing disconnect between content creators and AI-driven platforms, and why this shift is forcing the industry to rethink long-standing models of attribution, traffic and revenue.

The conversation explores how Tollbit is building infrastructure to help publishers regain control, enabling them to track, price and monetise their content in an AI-first internet. Toshit also shares insights on licensing frameworks, the evolving relationship between tech platforms and media companies, and why transparency and fair value exchange will be critical in the next phase of the internet.

If you want to understand how AI will impact publishers, who really owns content in an algorithm-driven world, and what sustainable monetisation could look like going forward—this episode is essential listening.

NOTE: This transcript is done by a machine. Human eyes have gone through the script but there might still be errors in some of the text, so please refer to the audio in case you need to clarify any part. If you want to get in touch regarding any feedback, you can drop us a message on feedback@thecore.in.

TRANSCRIPT

Vanita Kohli-Khandekar:Hello and welcome to the media room. Tollbit is a company that's a product of the world we live in which is about AI and short form and technology. Tollbit is a company that tries to monetise bot traffic for publishers and a publisher could be anywhere.

It could be a business standard, it could be a Times of India or it could be an Airbnb or an eBay. Anybody who publishes information on the internet is likely to have it stolen or scraped from the website and used in some other way and Tollbit tries to help monetise when AI bots come, they need to pay for that information. It tries to ensure that that payment happens.

Toshit Panigrahi and Olivia Jocelyn set up Tollbit in 2023. Recently, I sat down with Toshit Panigrahi who's one of the co-founders to understand what this company which has raised 31 million so far in different rounds of funding is doing and how it is helping the cause of publishers and whether this technology can be applied to video, to piracy or not. Over to Toshit.

Hi Toshit and welcome to the media room. Wonderful to have you here over from Boston. What's the weather like there?

Toshit Panigrahi: Amazing. Well, thank you for having me, Vanita. Weather is nice.

We had a very warm day yesterday so it's starting to feel like spring. I'm not sure if you can see but there's some snow outside. The last of it is melting.

Vanita Kohli-Khandekar: You and I had a chat a couple of weeks back and I wanted my viewers at the media room to know. Give us a sense of what Tollbit is about and how publishers are using it. Is it helping publishers to protect revenues or to protect themselves from bots taking away their data and therefore their content and therefore their business?

A quick overview of Tollbit and there it is.

Toshit Panigrahi: Yeah, so we like to say Tollbit's a platform that helps publishers of all sizes monitor, manage and then monetise AI bot traffic. And what we mean by that to really unpack it is our message to them was that in the future the primary visitor, the primary user of the web is going to be an autonomous visitor. Call it an AI bot, scraper, AI agent, AI application.

The words will change every year but the point is it's going to be an autonomous visitor. It's not a human visitor and while you might have many tools, many mechanisms to monetise human traffic to your site and you now think subscriptions and advertisements, affiliate links, you have no way to have a value exchange with this autonomous visitor. There's missing infrastructure and it's going to be your biggest visitor.

How do we help you prepare for that world? That's the infrastructure Tollbit is building.

Vanita Kohli-Khandekar: Okay and how do you monetise a bot? I mean how do you get the bot to pay? So you do a deal with the companies which are running the bot and you know if you could and also that whole thing of the Tollbit domain if you could just.

So if I was a publisher, let's say I'm the Times of India, I come to you. What do you do?

Toshit Panigrahi: Yeah so we have two sides on the Tollbit platform. One is for publishers and one is for AI companies and the goal of this is to make what we realised was that we have to make the flow of dollars for data happen faster or in this case rupees for data happen much faster because the way that the licencing deals, the content access deals, the data deals are being done is a very human-centric process. You have business development people who debate the price, you have lawyers who go back and forth on their contract terms and we realise that that just doesn't work for the AI era anymore.

We need a faster more programmatic way to do it and so what we did with Tollbit was as you highlighted when a publisher onboards under Tollbit, we give them first an analytics tool. We call it Tollbit Analytics. Tollbit Analytics is we like to say the Google Analytics for AI bot traffic.

So we are able to tell you who's visiting, how often they're visiting, what they're getting, what they're accessing, how many referrals they're setting back, are they abiding by the robots TXT and then we let you set upload custom licences. We have a set of standard licences. A publisher can upload their custom licence and then you can set rates against those licences and then we activate what we call our bot paywall.

So every site that onboards onto Tollbit gets a Tollbit subdomain and at that Tollbit subdomain we like to say you know www is where your human traffic goes and the Tollbit subdomain is where your bots go and that's where we create the agent paywall.

Vanita Kohli-Khandekar: And these the money which is charged to or the licences that you give to the AI companies which are running these programmes are linked to CPMs or cost per thousand that advertisers use regularly in digital advertising, correct?

Toshit Panigrahi: Yeah and that was very important. It was a core innovation. I think the industry takes it for granted today but when we first proposed that, that was actually quite groundbreaking, right?

My background, right, the prior three years before starting Tollbit, I was running an ad tech business at Toast, a point of sale company and what we realised was that in the future, you know, who was I going to show my ads to if one day agents could just order food for users? And publishers were facing a similar disruption and so what we realised was that it was a very interesting way to create this economy where we can tell a publisher that people are not visiting your website, every publisher knows what their CPM rates are and if this is a, you know, the way you should be thinking about this is replacement economics in a growing market, right? So humans might not be visiting your site, the bots are coming to your site in greater numbers so what if instead you could make money off of them on a CPM basis and then publishers could instantly understand what the economics of that future could look like.

Vanita Kohli-Khandekar: But does linking it or hinging it on CPM limit growth because a whole lot of bots also go to paywalls websites, they go to premium websites, how do you correct for that in the pricing mechanism? I don't know if you've reached that stage per se, you've got what 7,000 publishers currently?

Toshit Panigrahi: 7,000 publishers.

Vanita Kohli-Khandekar: About 20 in India?

Toshit Panigrahi: Yes.

Vanita Kohli-Khandekar: Yeah, so how do you correct for that because what let's say my paper business standard is behind a paywall, so anything I write is behind paywall but a CPM linked payment mechanism for bots might actually under index what they should be paying.

Toshit Panigrahi: Yeah, it's a fantastic question and one that we get I think early on from publishers. The reason that the toll bit analytics tool was so important was that because we could go to publishers and we could show them that, you know, your search engine referrals might be going down, your human traffic wouldn't be going down, but the one number, the one metric that keeps going up week after week is the amount of bot traffic that you're getting. People might not be visiting your site but they're asking AI applications the questions and so these AI bots are coming in greater and greater numbers and I think the sort of light bulb moment for a lot of publishers is when they realise that today monetization, web traffic on the internet is capped by human attention spans.

There's a limit to how much we can read in a day, right? There's a limit to how many articles we can read in a day but there's no limit to how many articles a bot can read when it's doing research and so as a result there's actually the amount of traffic on the web even though your human traffic is going down on a per site basis, the amount of bot traffic and page views on the web itself is about to explode in a way we have never seen and so this becomes, you access is going to be really high and the reason this works on a CPM basis is bots need to come and access that content tens of thousands if not hundreds of thousands of times a day because they need to load it into context to answer a question.

That's why the analytics tool is so powerful to illustrate to publishers that it is possible to charge microtransactions hundreds of thousands of times in order to essentially replace the revenue you might be losing from you know a zero click world for example.

Vanita Kohli-Khandekar: Some sense of how publishers are doing you know what kind of at what level is you know whether it's in payout terms or you know any specifics you can share or case studies or an example from one of your 7,000 publishers.

Toshit Panigrahi: Yeah we can tell you that that publishers make anywhere from the first thing that we tell a publisher when they're on board is we're not going to promise revenue right we're not going to promise anything on day one but we'll show you that this is the way the direction the direction the future is going and Tollbit actually becomes one of the more I think credible ways they can believe how a zero click world could operate.

The next thing that happens is you know we like to under promise and over deliver. I think a fair bit of publishers do get surprised when in the first month or two they might get a check from us right. It's not you know oftentimes it's not you know we say that it's not going to make your quarter but it actually shows that this programmatic licencing model this programmatic access for these AI agents and bots actually does work and so now I think out of that 7,000 I believe with at the end of 2025 we had paid out to about 20 percent of that of the network of sites on Tollbit in some capacity.

The payouts per month could range from hundreds of dollars to tens of thousands of dollars and so we actually see liquidity in the market and we can actually see what people are willing to pay for different types of content which is I think to me insanely fascinating.

Vanita Kohli-Khandekar: What does Tollbit make out of the whole thing? How is your revenue sort of protected?

Toshit Panigrahi: Great question. So we fundamentally want publishers to believe that we are on their side. We don't charge publishers for our services.

To date there's no product on Tollbit that we have charged publishers for and it's something powerful to ally ourselves with publishers because what we tell them is look we will make money when you make money and our investors are aligned with this. Our investors believe in this mission. We believe in that future where every you know we think in five years we will look back and we will think personally we will think I think that it will be insane that we ever debated whether a bot should be paying for should be identifying itself correctly can misrepresent itself as human or just has free reign to go scrape whatever it wants on the internet.

It will be insane in five years that this was allowed one day right and so what we you know the way we make money is we charge a transaction so publishers set their own CPM rates on the platform against whichever licence they want. Is it a full display licence? Is it a retrieval licence?

A summarization licence? And then we charge a transaction fee to AI companies on top of the price of that content and we do that in exchange for having done the business development work, the contracting work, ledgering this and making sure that they stay above board for accessing this content.

Vanita Kohli-Khandekar: Okay, correct. It's a question I'm I was curious about and I think we discussed it earlier. Does this work as far as piracy goes?

I mean I would I I'm very curious if entertainment companies have approached you or can toll bit work to help contain piracy?

Toshit Panigrahi: So it's interesting.

Vanita Kohli-Khandekar: What goes what gets lost to piracy?

Toshit Panigrahi: Yeah it's it's interesting that you talk about piracy right. One of the ways that that you know very early on right like there's a lot of these terms retrieval, AI agents, you know on-demand licencing that we take for grant CPM basis that we take for granted today. It was very difficult to explain to publishers in the beginning what exactly toll bit was doing because they've never seen something like this before right.

So an analogy that we drew is we said that piracy existed in music and the key to piracy right you can't catch all the pirates you can't block all the ways in which people can pirate the content. Instead you need to open up avenues where it's actually easier to pay for the content than it is to pirate the content.

Vanita Kohli-Khandekar: Correct.

Toshit Panigrahi: And that's what we focus on right. We're not a cyber security company because the people who want to scrape the content will find ways to scrape it and we consistently show that right. Instead of doing that let's give them ways because these AI companies one of the very very key learnings that we had very early on is people think scraping content is free.

Web scraping is not free. Web scraping actually costs money right and you know every quarter we release a state-of-the-bots report. In the last one we talked about 40 different companies that AI you know companies use in order to access this content right through web scraping means.

Our goal when we told publishers this was to say that we need to make sure that you know AI companies are paying for content and data they are just not paying you for the content and data. We need to open up a mechanism so that that money can flow to you not these third-party web scrapers.

Vanita Kohli-Khandekar: Right now you're talking about textual data. What about video, audio, has any work happened there? Has there been any demand there for a Dolby kind of product?

Toshit Panigrahi: Yeah so today our business we focus largely on digital publishers and to us you know a publisher we use that in the broadest broadest sense a publisher is anyone who has content or data that can be scraped. This could be a digital publisher sure like Business Standard it could be an e-com company it could be like for example my former employer Toast had online or had an online ordering platform that was scraped all the time. It could be a forum user-generated content site.

This is where we largely focus our efforts on. There's nothing that stops the technology from working with you know other modalities, audio, video. It just tends to be that things like video have trickier rights management considerations.

You know what happens if a logo shows up in there can I use this person's you know likeness in a AI output. So because there's a lot of open questions and thorniness there we tend to not work with those modalities just yet.

Vanita Kohli-Khandekar: Okay but you see yourself working with them in the future at some point when things sort of sort themselves out I'm assuming.

Toshit Panigrahi: Yeah.

Vanita Kohli-Khandekar: Okay you know this is a question I asked you earlier. The whole basis of the building of generative AI is that it scrapes data. It scrapes data from some of it may be just you know an Airbnb kind of site or something a Walmart or something but a lot of it is also original content produced let's say by journalists, by writers, by musicians and while there's a whole lot of copyright cases and regulatory debate happening the fact is on the search side AI has led to a drop in publisher traffic hugely across the world.

In India at least we've seen it fall between 30 and 40 percent over the last couple of years. How do we ensure that you know so this is a period of transition or transformation. How do we ensure that while this transition is happening the incentive to keep producing original content remains and I'm not sure if only CPM linked revenue payouts can help but I just wanted you to throw some sort of and maybe I'm coming to it as a journalist very likely but nevertheless I wanted your take on this.

Toshit Panigrahi: Yeah it's a you know one of the things that you're touching on is that at the root of toll bit is a fascinating sort of information theory and economics problem ultimately right. How do you incentivise the system encourage good behaviour for example and how do you incentivise the continued quality sorry generation of quality content because you know one thing is was clear from us from the start right which is that if you cannot monetise content on the internet incentive to create content dies and with it there's all sorts of follow-on risk to society right. Democracy dies with that right.

We can't ever afford to be in that world we have to get ahead of it that's why the mission was super important. Now in terms of what kinds of content get incentivised some of that actually lies in the data that we see of what people are paying for today for example right and I'll give you an example right one of the things that has become clear is if you have commoditised content right if you think about you know way a lot of a lot of publishers operate right where they're they're playing a game today but it's a game that is familiar to them it's an SEO game where they all try to outrank each other on certain keywords right and that works great for SEO perhaps but it actually hurts you in this new economy and I'll give you an example we know for and you know this in the in the most general terms right we know for example that 100 million people in the US will search for Donald Trump every day. If you're a publisher it would be that relies on on digital traffic you would be silly not to write three or four articles about Donald Trump every day because you know that it's a safe keyword that you can deploy resources against and recoup some and get some revenue from from right however if every publisher does that and every publisher writes four or five six articles about Trump every day it actually hurts your ability to monetise in this new agent internet because AI agents don't need 10 articles that say the same thing it's a wastage of tokens it's a wastage of context and you might not need to licence all 10 of those articles instead what we're seeing is that that the things that command a premium on toll bit right tend to be differentiated content because that's irreplaceable it's unique right they tend to be they could be premium content that's paywalled so if you're a paywalled provider the barrier to access goes up and so the what you can command goes up and this is why we see and freshness right so if you are if it's breaking news if it's fresh news then you can command a higher premium for it right so we're starting to already see in the liquidity that we have on the platform this start to shake out where you know local news commands a premium who would have thought it's actually the opposite of what happens on the human web local news commands a premium because it's it's highly irreplaceable content right you cannot get that anywhere else because no one else is writing about it right and so in that regard we see that you know in the long term publishers i think will adapt their content strategy just like they did with search just like they did with social in order to maximise what they can get from this sort of agent economy and my hope at least olivia and i our hope is that you know could this usher in an era of journalism where commoditised content is actually dinged and actually get new original content coming out of this but that's how the physical world works too

Vanita Kohli-Khandekar: and that's how paywalls were established they were established for non-commoditised content for premium content for original reportage or specialists etc so it is it is in many ways replicating what is happening in the physical world and what commands a premium in the physical

Toshit Panigrahi: world or the offline world if i may use that term absolutely now just now and but the goal is to do it at scale so that the bots can also access that right because the same way that the non commoditised content that for example business standard has is behind a premium paywall for for humans right imagine we're seeing that same effect start to net out where if it's a paywalled article which indicates it's hard to get to plus it's unique and non-commoditised it commands a high premium so you think academic journals you think premium you know are especially server paywall publishers they command a premium on Tobit you know you have the state of the board

Vanita Kohli-Khandekar:

report and i find that really fascinating i've had a look at two three one of them talks about real-time scraping i mean so we'll talk about real-time scraping how does it change the equation because when you're talking i mean does it is it that the frequency of scraping and you mentioned this in somewhere in the beginning that the volume and the velocity at which this whole happens is at another level how does real-time scraping impact how you price toll bit or how you

Toshit Panigrahi: look at the ecosystem it's a great question uh when we first started toll bit uh we went to every publisher in you know december 2023 q1 of 2024 and we told them every single and we told them hey we're going to help you get paid for your content by ai systems and we introduced them to this concept of retrieval of rag retrieval augmented generation output grounding different companies will call it different things but the concept is the same uh and uh we're not going to help you with training right we told them and it was a very unpopular position to take because every publisher was fixated on training and we said look we can't we're not going to help you with the training deals we're going to prepare you instead for this other world where bots will need to come in in in greater and greater numbers in order to access your content and that's how you create a recurring revenue stream because there's only a few companies that are that are training large language models and that might be a one-time check it's not going to be a sturdy recurring revenue stream you have to think about this as a new audience that will need access to your content and that number uh the amount of bots will scrape it in real time for retrieval is only going to increase now for the first two quarters of 2024 this was a very foreign concept to publishers and and honestly they didn't want to work with us because we weren't telling them what they wanted to hear but as we got them onboarded onto the analytics as they understood the architecture of ai as they understood what retrieval augmented generation was they started to understand what the market potential for this was the addressable market of monetising real-time access for retrieval is huge every single ai application is doing retrieval on the open web every single ai agent when you hear that word is retrieval none of that involves training and if that is the world's if that is the the internet's primary user in the future then you need to optimise and make sure that one your content is optimised for them and two you can actually charge them when they come to visit uh and that's how we architected tollbit tollbit is designed around that real-time access with the premise that the volume of access in the future is going to be far greater than what humans read on the web and we should be able to monetise that access on a cpm basis programmatically

Vanita Kohli-Khandekar: super um last question on next steps i mean you've raised 31 million so far uh 31 million correct am i right so far what are next steps are we looking at more capital and what is it more than the capital bit what do you see tollbit doing in the coming year month i don't know whatever

Toshit Panigrahi: forward-looking statements you can make so one of the things that we think about a lot is is uh you know the the core thesis of tollbit right is a is solving a monetization problem it's solving a internet economy problem right and if you think about where we started right uh remember when i said that i used to run ad tech at toast and i was worried one day that agents would order food and no one would look at my ads i think the world is now realising that this is not just a publisher problem right this affects the entire internet and this is why we architected told us that yes it works for publishers but it has to work for e-commerce sites it has to work for forum sites it has to work for you know b2b sas sites for example right every single uh vertical of the internet is getting disrupted right so um you know we're we continue to onboard publishers and that publisher growth stays strong but as we do that you know we're mindful that a publisher is not just a digital text publisher we have to make sure that you know the platform scales and works for the other verticals so from a business perspective that's what we're working on now uh in some of those other verticals uh and uh you know we you know again our investors are aligned here about what the the the future of this monetization looks like right the goal is not to to you know charge publishers it's to actually show that there is liquidity in the system that there is a thriving internet economy in a world where people might not be able to monetise via advertising anymore

Vanita Kohli-Khandekar: what's been the biggest challenge you're a young company two and a half years or so almost three years yeah you know three years at the end of this year what has been the biggest challenge to setting up toldbit and getting it accepted by publishers so one of the biggest

Toshit Panigrahi: challenges i would say was that like you said we're a small company we're a startup right and startups succeed because they find market opportunities to defeat the odds right we are creating a market where one one is the challenge of just point blank creating a market creating a market is difficult and we are trying to pry open a market where if you imagine the biggest tech companies on the planet don't want a market to exist because it introduces cost it introduces cost to something that they're just scraping today right without paying that uh and our sort of challenge was to really show that we're not antagonistic to ai companies that in the long run if you don't find a sustainable mechanism here that entire internet economy collapses and that's

Vanita Kohli-Khandekar: actually not good for ai at all we try we have slop at the end of the day all original content

Toshit Panigrahi: is gone what the hell are you going to be scraping absolutely right it's a net worse outcome it's for everyone right it's ai slop original content is not incentivised and the incentive to create any content just disappears right no one will share anything on the internet anymore right and that is a net negative outcome the hardest thing about toldbit has one is is is prying open a market when you know you have trillion dollar companies that want to make sure that you know probably doesn't form but in the long run is actually good for them and so we have to straddle the fine line we try not to antagonise ai companies we call out bad behaviour when we see it but our goal is to find ways where we can create positive value um you know there's a sort of narrative that that took hold right in 2025 that said we have to block bots we have to implement cyber security largely by the cyber security companies themselves and i think what we showed through the end of 2025 is that cyber security is a losing game if you do cyber security the only thing you incentivise is for the bots to get better at bypassing you right it doesn't actually solve the problem right you have to find you have to create the happy path and what we spend a lot of time doing is trying to figure out how do you create positive value for the ai companies because turns out they have many problems let's try to solve them if we can get website owners to cooperate imagine the things that you could unlock for ai companies with that that is i think the creative sort of exercise that toll bit does every day interesting i hope it works you know i find

Vanita Kohli-Khandekar: the bigger a company gets the more stingy it gets on cost so you have to train them to be a little more generous with what they spend on creating the product that they're trying to create you know whether it's apps or whatever with with the ai but wonderful nice talking to you toshar thank you so much for this and thank you for being on The Media Room.

Toshit Panigrahi: absolutely thank you for having me right now