The AI Revolution in Media & Social? - Part 2

CV:

Okay guys welcome to the second segment of AI in media and social. So just to recap last time we talked about AI in media and social. The quality quantity curve, we talked about modality we talked about personalisation et cetera. We will spend some more time on those topic, but we will also add a flavour-in of what do we mean by workflow depth, what kind of stakeholders we are building for, what should founders look at when they're looking to size up opportunities.

It’s funny to say it’s hard to pinpoint because I'm going us to pinpoint now. So let’s take a couple of stakeholders and sort of look under the hood. I know we talk a lot about work flow depth and we talk a lot about how there are some SaaS like characteristics that could be interesting for AI companies to look at building. So I think first let’s just sort of define what workflow depth means for us and then I’d like to open the OTT hold, I’d like to open the creator hold, I’d like to open the social media hold and look inside and figure out what kind of companies could emerge. So we keep mentioning work flow depth, a lot of people have raised their hands and said what do you mean by workflow depth, I think it’s useful for us to just unpack that and create parallels so that we can create a mental structure around the same.

Tarun:

So I’ll share at a high level maybe he can give more specific examples. I think in my mind we saw this playing out in enterprise SaaS also. Beyond a point every technology over time will get commoditized and the simple existence of that technology will not remain a differentiator. And we saw that literally especially in this space. You know, to the point to where we saw this company which was doing stuff around music, we’ve seen other companies do similar stuff. Over time what I think we all have realized is that this technology is cool but it’s not sufficient to create a large or to unlock large value. It’s a great utility but how do you build a business model around it is unclear. And I think we can borrow from what we’ve seen in enterprise software over there which is over time multiple things also become fairly commoditized there. Ticketing systems, ABC, what really makes it valuable is A, integrations and B, workflows. In my mind I think we will see some version of that play out over here. Companies/founders will need to identify their ideal customer, they will need to build deep workflow depth for the particular niche use cases. Locking themselves to the process of precreation, creation, post creation and in that there is value to be done. So I’ll give you an example, we haven’t announced the investment so I can't name it yet but we’re investing in a company that is helping creators create comic content. Now what’s the biggest issues in comic content today, the biggest issue in comic content is once you kind of visualize the characters and have a storyline it takes time to do a frame by frame screen sort of play of that and actually visualize each screen and then bring out the storyline. That is the number one bottleneck. Now if I can use generative AI and build tooling on top of these foundational models to say I’ve got creator X, I’ve got the storyline, just through prompt engineering can I basically just create almost real time every day episodes of my favorite comic content at scale. And so that in a friendly mobile TikTok style scrollable format. Now for that the technology by itself is not – it’s going to get commoditized. What’s going to be valuable is the tooling and the workflows which are relevant and specific for creators who are building this comic content Right, and I think whether it’s collaboration, whether it’s approval flows, whether it’s whatever like part of the editing process I think that is what will create value hopefully and I think there we will see more companies getting created.

CV:

Yeah, I completely agree. I think we were looking to just break down the creator value chain from let’s take a YouTube creator, storyboard the idea, shoot will take two days, then he’ll take another two days to sort of edit that, cut that, put the right thumbnails, put the right sort of edit chain and then publish it. When we say workflow depth in my mind I simplify that to how important is the tool to the creator or the stakeholder in their day to day routine. How much are they using it, how much of their life is dependent on it. So if we were imagining a situation where there is one tool that helps you manage all my library, all my content, it allows me to put subtitles for all my videos, it allows me to dub my videos into the languages that are relevant, it allows me to publish over a reel versus a YouTube shot versus an Instagram post and stuff like that. So not just production and editing but also distribution, dissemination and also management of a library. The more the number of boxes that you can tick mark in that as part of your workflow depth the more relevant and useful it is for the creator.

Tarun:

And then, sorry to lear in, and then intelligence such that there is a feedback loop which keeps helping you create better and better quality content based on actual real world live data in terms of how things are actually --

CV:

Absolutely. So the system literally should be able to look at your YouTube data and say this 20 minute video has generated better views than some of the others, here’s a quick ten short videos that you can put on top of this content we found engagement spiking at this second, let’s just put it out and it’s ready for you. Click to sort of post like. I think for me that exercise that we did internally helped me sort of understand how deep the workflow depth can actually be and how it’s different for each stakeholder. It’s different from enterprise versus a creator. I was actually wondering if there were other examples, so how would it look like for an OTT or a social media platform?

Aakash:

I think just to stick to the example Tarun picked up and because you said that this question comes up a lot that what does workflow depth mean I’ll just take two minutes to elaborate more on that with slightly more double click on it. Now that seemed comic use case, you can think of workflow horizontally, so that’s one. And the other is the technical depth, so the technical depth is basically saying that hey, you have found models which can do text to image. But can they do consistency of images which is what is needed when you're putting together multiple frames of a comic, they can't. First level of scaffolding which needs to be done, can they do control on posing of the characters, they cannot. You can't get that degree of control, that's the third scaffolding that someone is doing. Can they do multiple characters in a single frame and maintain all of this, so I think the investing company that we’re talking about that's a technical depth they’re building on that workflow, there’s a singular workflow of saying that hey, how do you get from text to comic. So one is this this depth, when you have to think of what all needs to happen to get to an output which matters. The other is exactly as you said pre-production, production, post-production, distribution.

CV:

That's useful mental framework, basically there’s a horizontal in which you do more things for them or there’s a vertical where you do stuff that is so incredibly hard that they need to be dependent on you. My inference then is what we don’t like is things that can get commoditized super soon. Right, it’s not horizontal enough, it’s not difficult enough a problem to solve for vertically and hence there’s not much value to be captured.

Aakash:

And then horizontal the other piece is not just horizontal enough, see, when you think of horizontal when you go down that value chain you have multiple stakeholders. Its like what we keep on saying single player versus multi player workflows. When you go horizontal you have to solve for multi-player workflows, it’s not just a single person you're collaborating with someone which is what from an enterprise SaaS lens Tarun alluded to. So it’s also that element in that if you build which is a multi-player and you build enough depth on the problems we can do you solve that's a compelling value creation.

CV:

Yeah, but let me pose the reverse question because a lot of these workflows are sort of dependent on foundational models and the big text in some form. I guess the elephant in the room is how much value will be left on the table. So it’ll be useful to just unpack that a bit. I know we’ve got our own points of view on core tech versus application Ed etc but I think it’s a useful distinction to have on what’s left on the table for me to sort of build on.

Tarun:

Yeah, so I think there better be enough because otherwise – so I look at this a little bit I guess just more first principles, right, I think in my mind what is if I were to just break AI it’s basically having literally large proprietary datasets, billions of gigabytes of data and then the ability to invest in advanced computing capability. The combination of these two is what allows one to sort of build AI models on top who has large proprietary data sets in government, who has potentially large resources to invest in advanced computer learning capabilities again sort of the larger ones. So I think there is an argument to be made that there is clearly a advantage to in governments. I think the big tech guys will look bigger five years from now, ten years from now, as a result of this. Just the way a lot of like Facebook was a web company even at the time of IPO. It became mobile first only after that and actually all the value creation came post. I think companies who are able to adapt to AI and take that platform shift and make that their first nature I think are the ones that will continue to do well. So I was just reading in fact today one of the reports that was circulated internally was on how Chinese companies are set up for capturing AI and Baidu is supposedly the frontrunner over there followed by obviously ByteDance and then Tencent and Alibaba. So my sense is there is definitely an argument to be made that when it comes to some of the more foundational stuff, when it comes to some of the more infrastructure like a cloud, all of that AI cloud I think that is an argument to be made that big tech will have an advantage. Thankfully I don’t think it will end there, I think there will be a lot of use cases niches within that. We give the example of comic content, we’ve invested in Murf which is doing text to speech. I think there are like hundred different examples where if we take each sliver of every industry whether it’s better at targeting, whether it’s better marketing efficiency, whether it’s better different sort of layers of content creation across different segments. I think there will be a lot of application layer companies that will get created with better tooling, better workflows which are suitable to that or extremely unique to that particular workflow I think those are the opportunities where most likely we will find opportunities. I don’t know, Cash, if you feel differently.

Aakash:

I broadly agree but I think again all of us have been entrepreneurs in the past and anecdotally every time this shift has happened it’s the same situation we’ve had.

CV:

Yes, it seems, yeah.

Aakash:

Yeah, always seems that the incumbents have the advantage.

CV:

Again IBM will have an advantage or a Xerox will have an advantage.

Tarun:

Actually that's a great point. There were mobile first companies created by startups as well, right, and so maybe we will see AI startups that will also come.

Aakash:

Like anecdotally you’ve seen enough but I would say maybe we’ll see new big tech companies also emerge and highly likely again historical evidence but yeah, broadly in agreement on saying that hey, there are enough value pools to capture. Work customer backwards, TD spoke about Murf, we earlier spoke about the other company that we’ve invested in which is doing web comics. These are things where you're going AI first and you're saying how can I use this as a point of leverage as a founder to generate create value and build an enterprise which can extract value because I'm creating value and it is meaningful enough. So founders are going to use this as a point of leverage and say all of these things, efficiency, creativity unlocked, hyper personalization. If these are the things at my disposal what can I do with it and solve customer problems, consumer problems, create new experiences which Tarun spoke about, there is enough room to create value.

CV:

Yeah, I echo that part. I think I'm more optimistic than the both of you guys are on new companies and new value creation. I think there are certain signals you want to look for if you're going to look at deep enough value pools. One of it could just be like we saw this example of I think the company is called Magnify.ai. They literally work very closely with sports broadcasters who’ve got a ton of IP and they help them create content out of that IP. All the historical IP, I want to see all off drives played by Sachin Tendulkar against off spin and then you could just extract all of that out. That allows for one me as a user maybe I want to see sport in a different way versus how you see it. And that is a big signal for me to say are there clear – are there these vertical applications that are still very large as a size of prize on sports is huge. And there’s a lot of proprietary data here that's not going to go through just the big tech. Somebody is going to have to come and build value on that. So I can clearly see vertical pools exist basis proprietary data, any other signals that are useful for an entrepreneur to look at to say maybe there’s something here, I should look under the hood and figure out if there’s more value.

Aakash:

I would say if you play to that democratization of pooling it’s limited purely by your own imagination. So if you especially go back to media and social as an entrepreneur just it’s limited by your imagination. You spoke about sports, we already have companies doing multiple other things. It’s almost like in gaming, in gaming we’ve seen. It’s not that gaming stops, right, or gaming company creation ever stopped. You have big incumbents, you have large publishers and yet every few years we keep creating bigger and larger gaming companies. It’s the same thing, I think media social you don’t have any restrictional limitation, it’s going to be limited by founders’ creativity and imagination of experiences, unlocking creativity, personalization. Just play to those vectors, go first principle at it I think there is enough to chase and build value.

Tarun:

So I’ll tell you what I like and generally the way I look at it is anything that so a step back right, when you're looking at a new space and you're asking so what’s going to create value I think the framework I like to apply to myself is to and maybe this comes a little bit because I also spend time on marketplaces. So I ask myself is there a 10x new experience for the customer that is getting enabled as a result of this. And we saw with marketplaces how that played out. Second, is there incremental value capture for the supplier which in this case would be the content creator. And value capture could be both in terms of dollars, it could be in terms of time, it could be in terms of greater earnings. So I think if I were to apply this and it’s a fairly simplistic model but when I look at these two things and then I apply what we earlier spoke about which is cost and time to create content is going to trend towards zero my sense is a trifactor of these three things is going to actually create new companies that at least to me that is what I would look for. Which is can I create like for example in the again I'm going back to the same example. I don’t think the company we invested in was a viable investment until generative AI and we’ll see how it works out. It’s still very early but for example without generative AI I don’t think there was an unlock possible where I could create new comic content every single day in a serialized fashion real time at a cost which made it economically viable where the customer was willing to pay for it. With generative AI I think it’s possible to do that where every morning I wake up just like people spend ten minutes on their social feeds I think if I'm someone who likes reading comic content every day before I sleep, when I wake up, whatever, I'm taking a lunch break I want to see fresh content every day that wasn't possible. I had to wait for one month till the new edition of that comic book or that Web toons or whatever came out that’s when I would be able to consume it. So to me that's an example of a new 10x consumer experience that gets unlocked as a result. And where I think even on the creator side of it there is an economic unlock that is happening. So I think I guess at the intersection of that is where I'm most excited.

CV:

That potentially also allows us to monetize differently because it means different for each user, you can potentially charge for that consumer surplus in some form so different kinds of monetization could also unlock.

Tarun:

Yeah, so I think those I think that's the piece of thing that I'm most excited by.

CV:

Got it. And if you were to sort of all wrap all these points together what are the things we like, what are the companies we like I think it’s also useful to sort of put out there the things that are interesting to us so how would you describe these are the kinds of companies I'm looking at, these are kind of segments that are really interesting.

Tarun:

For me it’s the same thing honestly, I don’t think that I’ve a different answer there. I think it is all of us I think we have to be careful not to invest in stuff that looks cool but will end up being a utility. And there’s lot of examples from Web1 and Web2 where stuff that -- actually but they just aren’t large businesses. And they’re great utilities at that. I think we will see even more of that happening in the AI world but I think what will create value is if you just again go back to his point on just customer backwards value creation first principles. I think is there a 10x unlock that is happening for the consumer and is there a 10x unlock for the creator happening in terms of economic value capture. For me that theme is the most exciting.

Aakash:

And with those two principles covered I would just add one more third principle especially for a founder perspective would be if you're a founder you should think hard about again the technical data and which goes back to some things become a utility or too democratized. If you're just a wrapper on top of – so I don’t know the answer to what for sure it’s more first principles.

Tarun:

If we knew we would be building it.

Aakash:

And what not for sure is where you're just a thin wrapper where it might seem super cool, it might seem that it is a multi-fold jump on the status quo but it does not have the technical depth. It is just a wrapper on top of foundational stuff, those things are going to be super hard. It’s not that they will not capture value, some of those are already capturing like there are people who’ve just built hundreds of tools, there are multiple websites where people are tacking these which are already capturing like hundred thousand dollars, four hundred thousand dollars monthly. But these are not long lasting, these are not durable. Because they’re just wrappers, So I think that is the other thing at least --

CV:

Yeah, so there is a real sense of depth that we’re looking at either vertically there’s a technical skillset or it’s just so like you said the multi-player strategy. There’s depth that the wrappers we feel will end up being boutique businesses potentially not for us. I think we’re running out of time and I know we can keep imagining and keep going on for hours on this but let’s end this session here. I’d like to just give a call out, if you're building this space of AI in general we do have a running list of companies that we track. Do write in to us and we’re happy to include you in that list, it’s just a globally viewable list. It’ll help other founders, people interested in this segment to reach out to you. Leave a comment, send us a mail, we’re just happy to include you. Also if you have interesting ideas and you're looking to startup in this space or if you just want to grab coffee and put on the imagination hats with us we’re happy to do that as well. Please feel free to reach out to us. Write in, comment, we’ll find you. Yeah, and that's it. Any closing notes?

Tarun:

No. Excited and I think for a lot of us who’ve been doing this now for some time this really, really seems like a serious delta shift in sort of the kind of companies and the kind of value creation.

CV:

Yeah, perfect. Thanks.

Tarun:

Thank you.

Aakash:

Thank you.

‍