Marketing to agents - Magic Episode 01 || #airpulse #generativeai

0:01 Hello everyone. Welcome to our episode five of Marketing to Agents. We had a lot of good news coming out from the market this week and we would love to bring that news over to you and just have a discussion around that. So again with me, I have my co host, Lalit, who's an industry analyst and good. Has a great vision on what things are going on and how things are panning out in future. Hey Lalit, welcome to the podcast.

0:29 Hey Ashwini, glad to be here as always.

0:32 Fantastic. All right, so we have a lot of, a lot of interesting news to cover today. So let me first start things off with what's happening with Google and the publishers. So I think we've covered this a few times in our, in our discussion where we've discussed how there is uk which is, you know, giving publishers a way to get out of all the AI searches. And then we're also hearing there are different ways in which publishers are signing up and making agreements with, for example, ChatGPT to have exclusive rights to their content.

1:08 This is something coming out of Google now where apparently the Google AI overview which we see in the Google search, that placement is going of course against publishers. And maybe not everyone knows about this, but Google does have a program for publishers where if Google is using Publishers content and showing that as news, Google pays them something every month. Now this time Google is trying to, you know, hand twist them, arm twist them into coming on those AI overviews, otherwise it will not make them part of any of its paid programs and use their content for training.

1:45 So what's your view on where is this going? Where is this LLM versus Publishers story going on? Like everything we hear sometimes they're collaborating. Now we are hearing that they are now trying to gain access to their exclusive content. Where do you see this going?

2:04 Yeah, so this is, so what we're discussing is being reported by the information and they have their own sources. So nothing is confirmed here. But it is about Google sort of allegedly launching a new AI news product. So Google News is definitely like one of the very popular destinations for discovering new news. And we always wanted it to be like, you know, should have a significant AI flavor where we don't, we don't have to sort of skim through a lot of links, but maybe news should be composed for the user.

2:38 Right. So I think it's about time that that product is on the horizon and the context is now that what is going to be the new relationship between the publishers and Google with this new news AI product and what is being reported here? Is that the publishers are expected to give broader content rights to Google. That enables Google to train their AI on the publisher's content.

3:09 Right. And they are making it as like a mandatory thing for publishers to get even for publishers to participate in this new AI news product, they have to like it is all bundled together that they, they have to bundle the content rights for Google AI to train on their content also. And there are also alleged rumors that they, this is like an exclusive partnership, et cetera. Right. So Google is basically trying to sort of leverage their high surface area.

3:42 You know, they have this distribution, right? They're one of the largest consumer distribution. They're now allegedly using that to basically strengthen their position in the AI pre training as well as the overall model space, right. We expect them to launch new consumer AI products but arm twisting publishers in giving them exclusive rights or rights to train content. These are two separate things, right. If you look at it from a consumer point of view, we deserve to know the news, right?

4:15 And we have trusted, you know, sources, or rather we trust some media brands for their unbiased reporting or whatever is our bias, right. We have our own preferences and Google has always maintained the position of a neutral intermediary. And today if they are sort of deviating from that position and trying to, you know, in, in their intent to strengthen their AI position, if they're trying to sort of deviate from their promise to the consumers as a neutral party, that is going to be a little sour from a consumer taste point of view, right?

4:52 And I think what, what it means for publishers is, is to basically start investing in their own AI products. That is very important. We all know why Google is doing this because you know, new information is what is always highly valuable compared to let's say old information, right? Because there's something called as a time value of information, right. And or half life of the information, something like that, right?

5:24 So what is the time that it takes for the information to become, let's say, not so critical? So that is like, you know, in somewhere on the Internet I read that it is, it's just two days. In two days the information becomes half as relevant because the domains are evolving quite fast, right? So the latest information is of very high importance and people who are people like Google are trying to use all their muscle, all their industry weight to get access to it in the best possible way.

5:57 So I think this should ideally give way for a more open industry standard where independent journalists, there is anyway a large volume of independent journalism that is happening on platforms like newsletter, platform it is YouTube as well as there are a bunch of newsletter platforms where like beehive, etc. Right. Where independent journalists are having their own subscriber base. So I think it will fuel that, let's say direction, it will act as a tailwind in that direction.

6:31 And I would also argue that much of it in the coming five years we should see a blockchain based solution for news, especially where individual sources have their own reputation on the blockchain. There are independent verifiers who can verify the facts on blockchain and everything is actually open for interpretive for people to build applications on. Right. So imagine there are let's say independent verifiers who look at something is published and every time the acclaim has been invalidated by the verifiers, it is all stored in the blockchain.

7:06 It is not lost in the ether. This is what exactly happens today, right. So how many times a source has been invalidated or you know, verified as false claims. Right. That should contribute to the reputation, a real time living reputation of any publisher. Right? That is what matters. And that once it is there, right. Publishers can then actually have an open market where high value reputation, high reputation publishers over the blockchain can give access at a premium to AI crawlers or anybody who wants to use their content.

7:40 So I think all of this is just going to act as tailwind in that direction. Let's see, these are my thoughts.

7:48 Yeah, yeah. In my view, I think everything hinges on solving for monetization or where does the money flow. So I think why this news itself is even controversial is because, just because I think Google already has a program which is, I think Google Showcase payments, which is where Google pays these publishers today. Also, if it is using their content in search or displaying their content, I think publishers or all of these media publishers, journalists would not have a problem if their data is used for search and then eventually offer some training and then eventually shown to the end user.

8:25 As long as they get compensated for it. Because eventually, see Google's AI news product would make sense. Unless it is like really having journalists on ground or crowdsourcing the news somehow. If it is still relying on people who are on ground or actually publishing their own content, I think then that compensation has to be clearly identified. Some layer of monetization has to be built in. I think once that is there, I think then it does make sense that everyone is getting compensated for the product eventually.

8:57 If the product is completely cutting off of any layer publisher, then it's just building in whichever way Google is thinking of building it then. Yes, to an extent. Then it does create that conflict eventually. But I think finally it's about how these publishers get composition or how eventually Google's product make that money and then pass it down to the publishers. I think that's where finally maybe to just boil down to but great point that you know, this would.

9:23 It's still an open field. And where do these highly curated, highly original content publishers, where do they go ahead and publish their content? Do they build their audience directly or do they even rely anymore on searches like Google's to be able to get their audience built? So that's a big question which will remain open till you see where each of those. The whole industry lies on this whole AI debate.

9:54 Yeah.

9:57 Fantastic. Moving forward, while we are on the term, while we're on the topic of search, something interesting happened at Meta and Meta has been in news for a lot of reasons, but we'll pick up the one which is related to the search. So what we've just learned is that Meta AI mode is also launching on Facebook search. So interestingly. Exactly. Or we're visualizing exactly the way Google search works today where if someone wants to search on Google, they go and Google gives an AI overview.

10:30 Facebook search might also do something similar. Now the interesting part is Facebook already has a lot of a different kind of data. It's not high intent search, but it has a lot of groups or a lot of local events listings reels which get published every day on Facebook. And of course Facebook is used by still by at least 3 billion people. So it's interesting to see Facebook taking that route to be able to capture audience and see if they're able to bring something unique to it.

11:00 But yeah, that was something interesting around search. What do you think about it?

11:06 Yeah, so I think there are still a lot of active communities on Facebook. The younger generation might have moved away like the Gen Z and the even younger generation, but I think there are very, a lot of B2B brands do still maintain a very active Facebook community. Right. And I was actually part of a lot of communities where a lot of valuable information was stored because it still is a very, you know, it's a dominant surface area for consumers.

11:40 Right. So I think this was, this was, this is a great move from Meta to sort of make all the like discussions that a user is, you know, like all the communities or groups that the future is part of and where discussions are happening. And if you just look at, if this thing comes to WhatsApp. Right. WhatsApp has a much more daily active user base and we are all, whether willingly or unwillingly, are part of so many groups.

12:11 And a lot of valuable information gets shared, exchanged in those WhatsApp groups. Particularly now today the search is very broken. If I'm part of a group in a way, permissions to act, permission to access that information is there. So I should be able to get a valuable unreserved for my search. And I'm more excited about this coming to WhatsApp than Facebook because it sort of relates to me and the kind of, let's say people that I deal with.

12:41 So today if you search on WhatsApp, it is basically going to give you a text match and it's going to list you list of conversations where let's say something matched which is, which is hardly useful, right. Especially if you are searching for an answer. Right. And if tomorrow there's a AI mode where you know, it is quietly indexing what has been discussed and it surfaces a really good answer. Imagine if someone is searching for that. I'm looking for a video editor across, you know, I'm part of like 20 entrepreneur groups and if I search, let's say you know, looking for a video editor or a video editing agency, right.

13:20 And if it can compose an answer which says agency X has been recommended in three groups by those so and so people. Right. This adds like tremendous weight, right. For me to prefer that. So I would say this is a. Is a really interesting dimension which meta has access to and gladly, they're working on it.

13:44 Absolutely, absolutely right. And then to think about it from a brand's perspective, I think for a lot of brands, like you said, they're still active on Facebook, but it's more to build an audience. And therefore I think now given that for newer players, sometimes getting into organic rankings is very hard just because a domain authority of an older brand is extremely well established rather than a newer brand. Facebook could give that opportunity because being active on Facebook, pushing a lot of content and this would include like you said, would be all Facebook properties.

14:18 So it could be Instagram as well. A lot of new brands, a lot of D2C brands pushing a lot of content on Instagram on Facebook. So I think if Facebook starts using all of that content and is able to get you right answers from search, I think that could become a very dominant pair. And because domain authority may not have such a huge significance there, I think they would still look at follower counts or interactions on engagement on the post as a count.

14:42 But another. So that could be a Great dimension. And then what I've noticed in my experience of using Facebook off late, there's also the whole Facebook marketplace and groups, a lot of that is very local. So I think this could also power a local. Enable a lot of AI local searches as well and could help in bringing and surfacing new content, new brands out of that. So as a brand I think Facebook has just brought itself back into relevance from a marketing point of view.

15:13 Not just Instagram but core Facebook algo as well.

15:16 So.

15:17 Yep, absolutely. And they have.

15:20 They might have to sort of navigate the their at least the impression of end to end encryption, especially in messaging apps. Large population still believes that it is end to end encrypted. So how do they index it? How let's say this AI mode comes to WhatsApp. It will. So it will be interesting to watch.

15:41 Mm yeah. And I now and again just tying it back to where we started at the Google. So I clearly remember that Meta also tried a news product, news aggregator product in different countries, in different locations just using that and it could not get that launch in Canada because it was not the. The somehow the discussion fell apart because the Canadian government or the media groups felt that the composition wasn't right to be able to give their access their content fragmentation.

16:15 The Meta Meta news product doesn't work in countries like Canada. So it again just could be great settings for how it plays out when it goes to AI searches and overviews. So yeah, just going back to that discussion. Okay, perfect. Moving on now, changing gears. Let's come back to a topic which is very which we are quite familiar with, AI visibility, Search and visibility.

16:45 So this week there was an interesting report published which showed that across. So this report analyzed a number of prompts and what they realized Is across top hundred prompts ran across all the four major platforms, AI search platforms, only 36 brands showed up consistently. Okay, so just to Repeat that only 36 brands were shown up or cited or were mentioned by AI searches across 100 top prompts across different categories too.

17:24 So what this tells is number one, it's an extremely, extremely narrow funnel right now. So it's only very few brands which are getting all the visibility from most of these AI platforms. Interestingly. So that does showcase that there's some problem with the AI visibility which brands are trying to build for themselves. Either they don't have proper tools ready, either they don't have their SEO and AEO programs fixed up or there are some brands which are not even making that right effort.

17:57 So that seems like a massive Problem but brewing up in the, in the industry. What do you think about it?

18:05 Yeah, I think it is pointing to the, the fragmented or fractured nature of the AI search. Internet has a lot of information sources and different AI engines have different preferences attached to it. For example, YouTube is a property of Google. So all Google products might give disproportionate, you know, maybe it has access to YouTube more than anybody else and then you know, it gives preference to it. So Google mentions or maybe YouTube mentions maybe result into visibility on Google Google AI products.

18:44 And then there are other brands which are slightly more neutral like Perplexity. Right. And then Claude, they have, they have different search backends as we earlier discussed that, you know the pre2022 era of search which was highly concentrated on Google, like 80 to 85% of searches used to happen on Google. Now that's not going to be the case moving forward.

19:07 Right.

19:08 So what this has an implication for the brands who want to optimize their AI visibility is that they have to, they don't have to, they have to make sure that no engine is left behind in terms of optimization because their target audience, unless they have a very good grip on what is the search behavior of their target audience, they cannot ignore any surface area. So monitoring tools, optimization across engines becomes all the more important.

19:41 Agree with you. I think monitoring starts off is again the starting point. So to be able to understand what prompts are relevant as a brand, as a publisher, it's important for us to understand which prompts are really important for us and then are we ranking those or not? And if not then there are corrective actions. It could be on site that things you have to fix or it could be off site where if there are things that you need more backlinks or you need to do more things around getting mentions.

20:13 So yeah, it's a combination. But looks like this is the whole AI search today is getting focused on very, very, very few players. These could be companies which are long established or have great already great SEOs but then getting newer brands to get visibility or get more traction. So I think AI focus, focus on AI visibility and ensuring that the tracking is in place is the starting point. And that takes us to another interesting news that came out this week was around a strategy which a lot of companies had been following off late.

20:53 So this strategy was that Listicles. So in fact in one of our earlier episodes also we've also talked about Listicles and how important apparently it was. I think ChatGPT or Gemini, one of those two LLMs which was giving huge weightages to Listicles. And this came out as a recommendation from a lot of AEO optimization tools where you know to write more listicles with your brand mentioned into it. What we heard today from a study that we see today is that in 69% of those cases where a Listicle is cited, the brand which is publishing that listicle is not even mentioned.

21:34 So it seems to be a very ironic that through Listicles you're actually educating the LLMs or the SC Google search about your competitors rather than yourself. So again there is jury is still out on where or what should a brand do? Should they not post, should they still post Listicles? But looks like Listicle could play as a double edged sword right now.

22:02 Yeah, so basically the intention of Listicle was to sort of get your brand mention or at least, at least get counted amongst the viable players in a certain category. But what it is acting as for AI search engine is, is like a, is like a, you know, like a citation for your competition. Right. So the self promotion in the Listicle that someone has posted on their own website, that is what is getting backfired.

22:33 Right. So which was always the, at least in the back of our mind we knew that it might actually happen because in a category, let's say if a lot of the new players are building these listicles so each of these listicles would obviously mention the existing players. Right. And the older the player more the number of, let's say dimensions. Right. So in a way it is acting as third party backlink or third party mention for brands other than the domain which is hosting it.

23:08 So I think what is clear now for marketers is that they have to sort of tone it down all the self promotion that they are doing via Listicles on their own website and sort of double down on third party or more authoritative third party publishers who can then give them brand mentions or brand citations.

23:36 Fantastic little point about third party and then reputation, good high reputation sources. So for example in your category, if there is a media house which publishes these Listicles or is talking about certain browsers, I think getting those mentions may be very highly valuable then brands writing about themselves and then just for the sake of it covering a few other known competitors in that place. Makes sense. Yeah, that could be a good mitigation for brands. Yeah, go back to your top industry publishers, media houses, content creators and then probably get listed with them.

24:14 I think that could be a good mitigation for that. Okay, sounds good. Moving forward we had then a news from Anthropic as well. So anthropic shift clawed Sonnet 5 as a new default for free and pro plans. So how does that change the world for our AI visibility or for brands who are working through that? What is your view on this change with Anthropic? Of course, I think it could be a way to also accelerate their usage and not for people to maybe go to open source models.

24:53 But what is your view on how does that change the world for the brands or the whole AI visibility domain?

25:02 Yeah, I think this one development which is Sonnet 5, right. Which is a highly capable agentic model. And what's the difference between an agentic model and let's say reasoning model? So obviously there's a little bit of reasoning in everything. But a model which is categorized as an agentic model is basically really good at long horizon tasks, right. It can basically stick to a task list which has been handed over to it.

25:35 And the list of tasks can actually include a lot of tool calls which generally come back and sit in the context window. So it has to be the reinforcement training on such models. It's very strict on, let's say not forgetting the task. As your context is getting populated with other things, other information tool called results, you not lose track of the task list. Such models are extremely good for your chief of staff use case.

26:07 These are extremely good used as orchestrators, right? So the parallel phenomena that is taking place is that lot of knowledge work is moving to AI agents, right. And almost everybody that I know of has a chief of staff which is either a hermes agent or OpenCloud agent. Right. And because of cost concerns, the main Orchestrator agent in each of these, let's say instances, whether It's Hermes or OpenClow, people have picked up a open source Chinese model or a cheaper model from Gemini as their orchestrator and they are, to be honest, they're not extremely happy with their choice.

26:51 But cost was always like the biggest, the biggest driver for that choice because agent tech work can be very, very long horizon and you know, they need some agent capabilities. So Sonnet 5 being available as an affordable option, right. With this new Anthropic release, I think this would make a lot of this a lot of these AI personal assistants in the form of OpenCloud, Hermes and few others. Right. They would. Consumers might make a switch to Sonnet 5 and when that happens, I think the likelihood of the discovery of brand and services happening over happening via Claude ecosystem, right.

27:37 That actually improves a lot. And so that, so the recommendation or the implication for brands which are Optimizing for the AI visibility is that they should not ignore cloud. They should go back to their robots or TXT and make sure that Claude is welcome and you know, guidelines which help them optimize their content for Claude specifically should also be implemented.

28:04 And does that from your, from what you just mentioned, does that also mean that as a brand I should focus now that this will drive more users to come to Claude eventually and thereby as a brand I should ensure that I optimize for Claude as well and like I prioritize at least optimizing for Claude because it might drive a lot of searches or search on the brand and in addition.

28:34 Yeah, go ahead.

28:36 Yeah, So I think many enterprises they have in their team plans, right? So there's always like a token budget and Sonnet being More token efficient. Sonnet 5 and cheaper as well. Right. So what, what used to. Just because of the, you know, like imagine if I'm an employee in a large company and I have, I have this overarching token budget or a overall dollar budget for my AI projects or AI usage.

29:11 Right. I am now quite open to continue to use my cloud for simple queries also because earlier I was worried about this search consuming, let's say more tokens and I might as well just move my search to a free option which is like a Google AI mode. But I can now just continue to work with Claude and ask, you know, let's say to do research and generate a report. Right. So what I've observed is that you know, Claude is being preferred for slightly more heavy duty knowledge work.

29:46 And search is not one of those heavy duty knowledge work. People sort of jump between a free or a dedicated search provider like a Perplexity or a Google AI mode and then copy paste the output from that to Claude for the next leg of the knowledge work which is slightly more heavy on reasoning etc. So I think with this people don't have to then worry about switching because it's more token efficient and budget efficient model.

30:20 That's a good point. And then because it's more, you know, budget friendly and so. But does that. I'm assuming that it would still have limits on how much free usage can happen. Right. Like today also if you're using cloud free or pro version there is a certain level of certain number of queries after which you are the cloud stops you. Right. So I would say it's the. I think Claude doesn't want to open that floodgate. It's just that you're getting a better model within those searches.

30:53 So I think that would be the case. Right,

30:57 yeah. So I'm. You're actually right. I think this news actually is from AI search point of view. It has a lot more, let's say much higher degree of impact on the search happening via personal assistants, which can use Sonnet 5 as an orchestrator and less as the surface area where users are searching themselves. But you know, obviously it's more cost efficient, so.

31:29 And that's why Anthropic has made it available in the free plan as well. So it's also a way for more people to get into the anthropic ecosystem. Right. Because if you listen to like the more vocal people, sorry, more vocal influences on the Internet, I think they are anthropic build. Right. So they are all. Everybody says so many good things about Anthropic and rightfully so because they're also extremely aggressive on pushing new things.

32:00 So I would say the consumer rub off that you get from, let's say, you know, like chatter in the knowledge knowledge work circles is bound to happen. And then when people come to Anthropic and they find good results and if they can use a good model in the free plan for a longer duration. Right. In fact they don't even have to upgrade if they can get the answers very regularly without restrictions, they would continue to use it as their main search, let's say go to destination and then that makes Anthropic a player that someone, almost every brand has to then optimize for.

32:41 Today, mostly brands which are B2B and deal with knowledge workers, developers, they are paying attention to Anthropic. But I do see Anthropic doing things like what we just discussed making a really good model as part of the free plan. These are all attempts in sort of expanding their surface area and bringing more consumers into it.

33:07 Right. I learned related to this there was something around Brave browser as well. Would you want to cover that too and just talk more about how does that lean into this too, this behavior.

33:20 Yeah. So one very interesting update also that happened in the last week is that on various search benchmarks, the brave context API, LLM context APIs, they've actually beaten ChatGPT, Gemini overview and Perplexity grounding. Right. So which means that now Brave is providing an equally good or state storage, state of the art LLM search infrastructure which is now available for anybody who's building AI search product to just plug into.

33:55 As we see the traditional players are becoming more restrictive, like people are pulling away their API access or keeping it very restrictive emerges as a viable alternative to basically rely on as a AI search backend infrastructure and offer a state of the art search experience for the consumers, especially people who are working with open source models. I think this can be a great, great addition to their jigsaw of tools which come together and deliver a great user experience.

34:34 So the implication for our audience is that they, they now have to make sure that, you know, Brave is also allowed as a bot to come and index and optimization of recommendations should also be then honored.

34:48 Fantastic. And then that would also imply that there are searches happening now if Brave, there are more searches happening on Brave browser itself and that in turn triggers.

35:03 No, it would not mean that Brave searches. So searches are happening on the Brave browser. This is an independent thing. The LLM context API by Brave is basically a search API, right? So that comes, that is powered by the Brave search index. And people can directly plug that into their applications or agents as a search tool. Just like there are publicity APIs for grounding into web search. Google Gemini by default grounds that with Google Search.

35:36 So I think Brave has been parallelly building their index. Now this is a, the update tells us that their LLM context API, which is there in a way, their answer API, it is on accuracy and on speed. It is state of the art now.

35:57 And today just double clicking on that Brave itself. How does that, how does that monetize today?

36:07 So APIs are obviously paid and Brave had a very interesting model. In fact, 2018, 2017 is when brave was emerged on the horizon. You know, in the public consciousness they actually came up with this concept called attention tokens. So Brave launched with the promise of, you know, you will never see ads on Brave, right. So it is equally fast, as good as Chrome and no ads.

36:40 And Internet monetization has gone to the, to the worst possible state. Where if today you go to a typical web news website like Hindustan Times or Times of India, the actual screen real estate that news has is less than 10% and 90% is what you see on your screen is ads, right? Clamoring for your attention. And inadvertently you end up clicking on an ad and which is like a false monetization for the, you know, bad experience for the user, false monetization for the, you know, for the, for the media brand.

37:15 And, you know, it's a fraudulent click for the advertiser. Right. So Brave was basically, and this was a state even back in 2017 also. Right. The Internet experience for media sites have been bad for last like 15 years. People are trying to juice every last drop of monetization from the web visitor that it is visiting. Right. So Brave had a great promise that they would basically not show you ads. Their front end would take away all the ads and give you a very clean user experience.

37:46 And as you use Brave, you start accumulating attention tokens as a consumer. Right. And you can have your crypto wallet and all of it goes into the crypto wallet. All your attention tokens are in your wallet. And brands, also brands which are where ads are getting stripped away. Brave is going to transparently build a wallet for that domain, like for example Times of India. So it would automatically build a wallet for Times of India, which Times of India business managers can come and claim and they will basically compensate the removal of advertisements from the user experience.

38:28 Compensate with let's say attention tokens. Right. So it was a great way for people to you know, like have a great user experience and also have publisher monetization. And what would be the value of these attention tokens? These attention tokens would then be used by, on an open market by, by brands to basically have specific targeting in Brave itself. Right. So Brave would then have search experiences or consumer experiences where ads would be lot more intent based and not intrusive.

38:59 And on those ad products, Brave attention tokens would be the currency. So this was the very, I would say noble intent but I think they got into a lot of privacy issues and hence, you know, it did not, it did not ever make it to the consumer limelight that it should have. But I think they are making a comeback with products for AI agents, especially search.

39:26 Interesting. Interesting. Yeah. The way you described its genesis, it was very interesting. And then I think if that would have worked, that would have been an interesting model to just look at how this commerce has been enabled and how is that attention tokens would have had real monetary value attached to it. I think a great way to see it. Perfect. That's most of it.

39:56 What we had from the news for this week. Anything else that you've come across which was exciting, which was interesting and very relevant to what we are talking about about here?

40:08 Yeah, I think one other thing that we discussed in our previous episode also which was this two new standards which Google DeepMind and Google Cloud Team published. One was this open OKF which is open Knowledge framework, which is, which is a great way for organizations internal knowledge to be, to be documented for LLMs and AI models. And it also, let's say if every department and every organization sticks to that format then it sort of enables the interoperability of internal Knowledge or knowledge per se in the knowledge economy that I think I'm very bullish on and I'm seeing on Twitter more and more developers, companies are adopting that.

40:59 So very soon we'll start to see that being very important. Second is, I think there is recently something which the, the, the, the, the, the W3C consortium, right? So they have now basically added another method on the HTTP protocol. So HTTP protocol which allows web to be browsed across browsers, right. They've always had like a fixed set of methods like get, put, post, right, Delete.

41:34 So they have now added a fifth method which is query, right. So on. So this is basically an acknowledgment that you know, websites can directly, you know, give an answer to a very conversational query. Now that query could be, could also be a read only query, it could also be a write query. But that I think is a big welcome change. I'm yet to go deep into it, but I think this was anticipated. So in future what will happen is that you can basically hit a URL of sense.

42:11 There could be a simple user interface where users can directly, let's say interact with the website, get answers without any intermediary.

42:21 Right.

42:21 And to make it very simple, the HTTP body has basically included this new method in the protocol.

42:32 Okay, interesting. So there will be. So I think, as you said, I think there, there are still more things unfolding on this, on this topic as like every week. So I think this will be something to keep monitoring and keep understanding what's happening here, right?

42:49 Absolutely.

42:50 Fantastic.

42:52 Basically when you hit a URL and you pass a payload to a request, HTTP is basically request, what should the server interpret this as the get method when you pass on the payload. And if the method is get, then the server interprets that this is a read only thing. It will never write anything on the server, right. It will not process the inputs to manipulate the data or you know, update its state. It is just a get query, right?

43:23 Post is basically where you are trying to update something, right? So basically when you're signing up on something, so you're posting the data, so the data has to be processed and you know, the server can recognize this as the user intent to update something or post something, then delete is another method, right? And I think query is, Query is another method which then has to be inter, it has to be understood by the website backend as let's say a query that has to be processed.

43:51 It's a conversation query from the user. So that I think first of all it's an acknowledgment of that the conversation interface is the future. And then websites also have to accommodate how to handle such queries and as much as possible, get rid of the intermediaries. Right? So you can basically host your own agent under your own domain, which will process all the, you know, queries, you know, the query method requests that are getting from the user.

44:23 So. Very interesting.

44:25 Very interesting. Yeah, indeed. Perfect. I think that. Okay, perfect. Yeah. That was great discussion today, everyone, and please do share your feedback with us. Love to hear from you. Thank you again for making it to the session.

44:45 Thanks, Ishmael.

44:46 Thank you. Bye.