How AI Agents Search, Read and Buy, and What Brands Should Change

0:02All right. Hello everyone. Welcome to the first episode of marketing to agents. As we all know, I think agents are becoming one of the biggest will become the biggest consumers on internet very very soon. And this is a show that we're starting just to understand what are the different aspects of marketing or how will these agents behave once they are online. And yep as always I have Lalit who's a thought leader in this space and hey lit great to have you here and lovely to have this discussion today together.

0:35Thanks thanks Ash and this will be our recurring feature. So really looking forward to having great discussions around how the space of internet is changing, how marketing itself is changing. just to start I just came across this very interesting fact today that the entire AI agent traffic right which includes the the crawling which is for indexing and then maybe inference time crawl and also on behalf of a user if an agent is coming to a website. All of that bucketed together under the label of AI traffic

1:12That is supposed to topple the whole whole human traffic on the internet by the end of 2026.

1:20So yeah as of June it stands at 80% of the human traffic.

1:26Wow.

1:27Very interesting. So that means

1:29Today there are more or eventually there are more crawlers who are using the internet than humans themselves. Absolutely perfect.

1:38Perfect. Well, so I think that leads to one of the points that we were earlier discussing. So that means if the internet earlier was designed for humans, the whole concept of a website for example, buttons to be clicked, if the primary user is going to be an is going to be an agent now, I think the whole paradigm of how the information is presented, how websites are designed will also change with that I'm assuming. Right.

2:03Absolutely. I think u the whole the whole funnel right right from discovery in of information which is which is we which is what we typically call as search right right from that to how people publish information all of that has to change right because agents do not have the limitations which human humans had right as humans we can when we search we we probably are you know it's not very not not everybody's an expert in defining exactly what they're looking for, right? So,

2:37Oh yes,

2:37That's one. And then the search SER page, a search engine rank page, right? That used to be like the golden real estate. And if you are on the SER page one, then there's a chance of you getting some traffic or business in that category and because humans would never go past the first page, right? Very rarely would go to let's say page two, right? So

2:59And also the entire judgment of what is the information is it good quality or bad quality other than the search engines humans had to go through the information and you know make the judgment for for themselves. AI agents are actually a different beast right so first of all they don't they don't have any limitation on let's say how many let's say links that they can consume parallelly they actually can go up to 200 links right and they can basically summarize the information in a much faster way so that is that is actually a different as I said it's it's a different entity on the internet so how we discover information how we consume information what do we expect from information is always going to change.

3:43Absolutely. I completely agree and I'm just this gives me an this just created a thought for me that how was it when probably when the first search crawlers came into picture like what was there at that time of course internet was much smaller or the users on internet were too less and I think that time the search crawlers would have been one of the key customers and all the websites were also optimized for them. So I think just to your point because these users are going to be different in their behavior the way they're going to consume information. I think a similar effort will a very similar effort from a lot of businesses or people who are presenting their information on internet will have to be done where it's more u consumable by these users.

4:32Absolutely.

4:33And you know as we speak it has been a few years into this domain itself right. So

4:39So and there have been let's say a lot of investments from major engines. So today charge GPT still is a market leader when it comes to the AI search but their market share is being challenged by Google again.

4:57Oh yes. So, so the the market share of overall AI search the charge share charge share is is right now at around 81% 81.4%. And it was actually 89% a year ago and then from there it has declined. Right? So there's a lot of focus on the the search or or rather let's say there's a technical term called grounding the information because these models are trained with a knowledge cutoff data right

5:33And any anything consequential that they are doing for the user right or for the enterprise

5:39It has to access real-time information it has to be grounded in actual facts right

5:44So search search is a is a very core more primitive for agents. So from that perspective also all AI major all major AI labs and major AI search players are have been constantly innovating and heavily investing into making the search a lot more robust

6:01And and the space has been very dynamic.

6:04Yeah. So what we what we knew as let's say conventions and best practices that used to work in 2023 on chart GPT where it was just a simple rag right where they would make the model would actually call a web search tool and then you know summarize the top top five results right from that point in time to now today when we are looking at search itself is getting executed as as code right which which has its own dedicated sandbox. Very recently, just yesterday or just in this week itself, Perplexity launched search as code, right? where the the the search intent of the user will be understood by the agent the search agent and then it would it would basically write the code which would which would basically you know have lot more control on the search not the tool but let's say the search index right like for that query what sources have to be ranked how many sources have to be obtained what should be the filter right

7:14Right

7:14And that output you know from that search as a code is then what is is what is given to the the user for rendering

7:23So the whole paradigm has actually changed

7:26So now that you mentioned so you mentioned a couple of things which which I would love to double click on so let's talk about the last piece which you mentioned about regarding perplexity so as a business owner or as someone who has a website who gets a lot of organic traffic on their website. What's an implication for them? Because if the way Proplexity is now handling search, that means every search or every crawl by Proplexity bot would be very different or would be very contextual to the search. So what does that imply for me if I'm a business owner and I have my own website? Like how do you see that getting impacted or how should I prepare for that sort of a future or that sort of a near future? Yeah, that's a very important question. I think across the engines whether it is judge, Google with the Google AI overviews and AI mode and perplexity. What is very evident is that u the output is becoming very much citationled. What is meant by citation le is that that the information that the websites should have or the publishers in general should have it should be unique. It should not be a secondhand or a third-hand information because

8:45Because the crawlers would get to know that is this piece of information unique enough or has that been let's say you know repurposed or reformatted from another source because they are getting they're getting to crawl the entire internet right so they want to site the original source as much as possible that's one second the search as a as a primitive is going to become lot more personalized, right? So, it would not be expected from the user to be ready with a very extensive prompt. It would be expected from the agent to understand the intent based upon memory or based upon let's say the you know the rest of the conversation. And the agent would make a lot of implicit assumptions about what the user's intent is. And what this would result into is the fan out queries will be actually long tail, right? It would have you know somewhere between five to six even more keywords

9:43Right so it would not be let's say like a very short keyword but it would be a longtail keyword

9:49Yeah we're already seeing data coming out that a typical query right now this is not of course this is user query is at least 15 keywords when they do it in chat GPT or in but yeah continue yeah

10:02Yeah so as a publisher of content I think first is that we have to u first of all make sure that the the information is technically accessible by the agents. So agents would highly value simple information which is presented in let's say easily easy to read not JavaScript or let's say not hidden behind or you know it's images not much media heavy in fact there will there is there is this concept called token budget right so every website which makes it to the short list of the search tool it is given let's say certain token budget and anything that the page contains which exceeds the token budget doesn't make it to the context window of the summarizer. Right? So it is very important for the publishers to basically create the most important most valuable piece of the web page. It should be above the fold on in the first fold and it should be it should be extractable. It should be extractable as in it should be encapsulated in a in form of a question and an answer or it should be self-contained with let's say required you know metadata and obviously should not have too many

11:22Extra tags which are unnecessary for the agent right

11:26And then as much as possible it has to be it has to be contributing something unique which the agent can easily site this website to right

11:37I And because you started off this on you started off this response by saying that because it's going to be contextual to that user. So I would also put somewhere that every business or every website should also start focusing a lot more on their persona or who's their ideal user because the more they emphasize that or the more they talk about that I think that way somewhere it'll be a signal because when those longtail queries are formed and then they come to you if they see that the the match is quite instant or the your answers the probability of your websites or your answers getting picked may be much higher. So I think that focus can also be a good help for the businesses.

12:16Absolutely. So in another discussion I actually gave this example and this very very very relatable example. Let's say you send an Instacart delivery person

12:31Or or rather an Instacart. You order something on Instacart and you're not very specific about it, right? And and let's say you're looking for a type of milk, right? You just said milk in your in your cart and it did not specify whether it is almond milk or it is let's say organic milk right and the Instagram fetcher is basically in the aisle and it is looking at let's say at least 50 options right now if the the the two things which are very important first is the in your input to the Instagram let's say request if if it was more specific the job of the fetcher would be much more easy. Right? Second thing is despite your instruction to the instammart fetch fetcher even if you have been very explicit about what you want but if the products out there did not have proper labeling right for example although the the milk is organic but it does not talk about or does not explicitly say that it is organic it would be very difficult it would still be confusing for the fetcher. Yeah.

13:39Right. So what this what this example emphasizes is that as publishers we are like the you know the the carton of milk sitting in the aisle right so the agent has reached the aisle which means we have made it to the let's say the the 200 set of links which which is then supposed to be analyzed for composing the answer and we have to be we have to be very clear about what we are right if we are an organic milk then we have to explicitly say on our packaging that we are organic, right? If there's any other benefit any other differentiator that we have, for example, it's a organic and let's say it's a specific animal milk, let's say goat milk, right? And if if the memory of the agent tells us that you know this person the end user would would really appreciate this, the your chances of making it to the card for the user is is very very high, right? So I think extraction of all the right information about your product, keeping your ideal customer profile in mind, having this very diligent extraction process, extraction process of the internal knowledge, the product information from the point of view of the customer and then making sure that it is very clearly written on the website and accessible for the agent is

14:57Very important. That way the chances of you making making it to the longtail queries for the end user when the agent is visiting your website is much much higher.

15:09Right. Agreed. Agreed. And I have a funny anecdote on what you just on the example you took. So I was speaking to a friend this week and [clears throat] so this guy took about 2 hours and he started his so he wanted to buy new sports shoes and he started his search on Gemini and he bought it like to the very last extent he bought it from Gemini like almost from Gemini. So his search started from comparing three brands and what are different types of shoes. It kept going. It kept going. Finally bought to a point where Gemini pointed him to a store where he can go get that shoes in next one hour at the at the most reasonable price in the whole region. So it was to an extent. So and of course Gemini being Gemini, Google could actually even get the inventory datas. It could tell okay which is which which store is carrying how much inventory of that product and then which store has it available at what price. So to your point exactly the more the more data or the more marketable right data we can present I think the more easier it is for these agents to find and then give users a very pointed answers because if the user is running their full journey full buying journey on an LLM or or with an interaction with a

16:28With a GPT this would be very critical to have every data point possible and available but

16:35Absolutely One final thought on this point before before we switch topics, but what if I'm a business and I try to be extra smart and I just fill my website or the first fold with all the right all the information like I overload it with information. So to your milk example, let's say I I put them I put organic, I put you know non-fat or lowfat. I put everything which could be just right to my persona but also a lot more information there so that oh just to you be greedy and oh I can be figured in 10 different categories. What what harm you think can that do to my website or or my ab ability for crawlers to come to me?

17:19Is that a downside to that? Yeah,

17:21Absolutely. I think so this is a very very important question and that is and this points towards the credibility of the publisher right or the information source and especially as a business right businesses are not operating in a silo and especially in 2026 where internet has a lot of us usage related content a lot of communities which are which are very authentically sharing information about let's say a particular brand right so the consistency the consistency with which the brand is mentioned right and the correlations right with which users are making that brand association with let's say what use cases what features right so the the times the more number of times the crawler actually comes across your brand in association with those keywords it is going to try to make make a very strong let's say indexation right with that with that topic right

18:18And and over and above I think if if if a if a particular brand is making an inauthentic claim right which cannot be packed up with let's say a certification which can be verified right it will it will come up you know you might actually fool let's say AI engines for the first time if it's a very very new domain but the moment let's say real users who have a bad experience right they come and share it on let's say Reddit or other social media right in fact YouTube is a very important source of citations right close to 25% you citations across on an average right across all these engines are are coming from YouTube right and that is where a lot of people like like us right they they share opinion about products they they sort of vouch for products so all of that is going to contribute to the credibility of a business understood understood so what so what I'm hearing you say is that u doing it too much would be detrimental because then your then that means your correlation with certain categories automatically becomes weaker because you just spread yourself too wide. If you are very relevant for let's say two categories, focus on those two rather than you know spreading it yourself across 10 different categories because then there

19:36Is higher chances that you might LLM might not give you weightage for all the 10 or will think you a lower preference for all the 10.

19:43Yeah. Yeah. So

19:44I think I think one very easy mental model would be to go ahead with the assumption that AI engines would deeply understand the user and they would want to deeply understand your business. Right? So in that attempt to make the AI deeply understand your business, you should not be let's say you know like fake in terms of let's say publishing information which is not truthful for the business. Right? For example, if a business is not let's say compliant on a certain standard, they should not pretend that they are compliant by publishing marketing content which which cannot really stand the let's say test of time or let's say which cannot really back it up you know as a claim whatever is the state of the business right it has to truthfully describe itself in the in the depth which is which is best possible right for the business so that the AI engines can deeply understand right so that that would be the way to, you know, think about things and move forward.

20:45Perfect. Perfect. Lith, U. Anything else? I think, I know we've, we've had a good chat here. but there were a few great headlines coming out this week. and anything that you would like to, you think we should double click for today? something around you did mention about charge GPT's traffic share of AI referral traffic dropping to 81% from somewhere around 90% or 89%. Who's the one that you think is picking up and what do you think that tells about the user preference? So is it because the models other models are getting better or people have had too much of chat GPD and they just want more variety? Is it different use cases? What do you think is causing that shift to happen? like is it just different services getting better but what what's your thoughts around that?

21:35Yeah. So I think two things that are happening which is sort of taking away market share from charge. First is I think Google is making a comeback right I they have actually done a good job in terms of improving their models with Gemini. the latest Gemini models are actually state-of-the-art and they they deliver a very satisfactory experience right so that is one reason that you know the the Google obviously has a the largest distribution right and them doing even slight bit in terms of making AI front and center would obviously take away market share right the second thing is enterprise enterprise adoption of claude right so for just B2B specific queries the claude market share has actually tripled right. So on or all the for all the enterprise originating queries claude is basically growing its market share. So that is that is causing a decline and for the consumer queries I think Google is actually increasing their market share. So that is causing a decline.

22:40Interesting. And I read I saw a tweet storm sometime this week. So there is a company called Lindy and they completely moved away from Anthropic this week to everything to Deepseek just because the cost of using anthropic was getting so high they've moved completely away shifted all their stack from anthropic to deepse of course validated it if they really done it or not but there was a tweet from one of the company leaders that this was a huge move and they've been doing they've been at it for a couple of months but now the move is complete and they've saved millions on just their spends.

23:18So I think and and they vouch that while of course Enthropic is better but DeepSync is almost 95% there with their V4 model. So to to your point just to add to that I think of course the models are getting better and anyone who has better distribution in this case of course Google they are able to of course surface it at multiple locations and and multiple touch points for the user and yeah and Google did another interesting thing I think which which I was discussing that they are now also using this opportunity that they have the users they have the agentic layer Now they are bringing it very closer to the commerce. So they've already been building Google shopping but now I think they are trying to make that front and center in agentic commerce space as well. Y

24:08I know if you like have some thoughts on that what you

24:12Absolutely I think what Google is trying to do with the the universal commerce protocol is actually a very strong indicator on how you know on how things are going to be in future right so one thing they've launched in fact they've announced in Google IO 2026 is something called as the universal cart now what this universal cart is is that you know across all Google properties which happen to be like quite significant in in a consumer's journey whether it is Google search, YouTube or let's say if you are on let's say Google maps also right so so you can discover products and you can you can then add those products to a a metal level cart when I say metal level this is a cart at the Google account level across merchants right it's not a merchant specific cart it's basically a card associated to your Google account and you can keep adding those products and then your Google payments right your Google payment will then for all merchants which are compliant with the universal you know the agent payment protocol the AP2 AP2 standard right all merchants which are compliant and Shopify I think is a partner there Google will be able to make the you know

25:30Payment on the users's behalf those agents will be able to make the payments right so that is a is a fantastic way to sort of give a great user experience where I don't have to really remember like who the merchant was the moment I discovered the product I can just you know add to cart and you know and I can also set the instructions that I will buy this if the price is let's say lower than a certain amount right so all of those things is also possible

25:55Wow right so that means that whole like last last 10 years or last 15 years the whole the whole industry for optimizing the shopping cart or optimizing the behavior from shopping cart to buying journey will now have to be rewired for an agentic shopping experience then. So how do you ensure that the agent gets that information knows all the right metadata or all the key highlights of that product and then ensures that it's competitive from other other merchants and then allows the buying to happen in the most seamless manner. So I think that whole journey will have to be rewired. Of course the website will remain but then there has to be a backend layer which rewires it for the agents as well. So that will be a big opportunity area as well looks like now.

26:45Yeah absolutely. I think all these merchants would have to pay extra attention and rather special attention to the product feed which which let's say Google shopping consumes or charge consumes right so the feed has to be very very detailed so that you know connecting to our earlier point of deeper understanding about your product right so the feed has to enable very deep understanding for the AI engine for the product so that it can actually surface it for the right user right where the conversion is much higher, right? And then all these merchants have to upgrade their payment infrastructure to support the agent payment protocol which is also called universal commerce protocol. I think there are like bunch of names which are sort of synonyms.

27:32So so I think these two things if if if e-commerce merchants can you know can sort of ready their stack for this they can actually you know ride the wave.

27:44Interesting. And now that gives me just one last thought on this is that as a brand either I really become a brand which a consumer remembers and specifically asks for it like let's say I want on shoes or I want Nike shoes for example or if I'm a brand which is not yet so consumer recogni recognizable then I should anchor very strongly on maybe one or two values. So it could be let's say the most comfortable shoes or the most best shoes for running or the most economical option for running shoes. So something like that I think because it has to really now very strongly anchor. So visuals actually that's another question u I'll probably ask you now how big a role will visuals play in in an agentic commerce you we we've seen all e-commerce sites right or let's say D2C brands highly strong on visuals great models great photography great videos now how does it entail in this whole agentic commerce does all of that carry any weight now so if I have a I have a brand I have a sunglasses brand and I've created great visuals, great photography, great videos around it. Does that even help my funnel if it's agentic commerce now?

29:03Yes. So there are two surface areas that we're talking about, right? So first surface area is where users are consuming content which could be social or agentic curation but that's like you know like a newsfeed kind of an experience where people want to you know consume content right

29:21That surface area is is obviously not where commerce happened with very high intent because I'm not going to that that that product or destination to do actual purchase right

29:33So that's where the brand gets built

29:35Exactly that is where the brand gets built. That is where let's say the affinity can be built. You can your visuals can basically establish an emotion connect with the right audience, right? And and that sort of builds a recall also, right? and then the the other surface area is your your chargity or your AI agent, right? Your could be your personal Hermes agent or your open claw where you can say that you know or it can automatically you know plan for you. For example, let's say there is a hike which is scheduled in the next month and in some previous conversations I might have said that I don't have proper hiking shoes, right? So the agent can maybe proactively you know discover that there is a hike and then maybe at least two weeks two weeks ahead in time I should be able to you know get a trial for some shoes. So it can then surface it up. So that surface area is where the visuals are not going to matter too much. But at the end of the day as a user when I am about to make the final choice if a certain brand has been successfully targeting me or let's say engaging me through the visuals on the first surface area which is your news feed right it it might have a much higher chance in terms of the recall in terms of the affinity to go for that

30:50Fantastic yeah massively changing paradigms and I think it'll be more important so feed becomes very important but at the same time yeah feed becomes important for brand building and then the agents and optimize for the agent conversions.

31:06Absolutely.

31:07Fantastic. And just your hike hike example, one more small point on that. So let's say would so let's say it's a hike that I'm planning to go let's say next month it's a famous hike would would the system with the agents get to with the agent gets to a smartness level where they look at the photographs of all the users publicly available for that hike. see what they are wearing, what gear they are wearing, recognizes those gears or tags those gears and then goes to those brands and gives me those options. Can the agent So I'm just thinking that how does that content get utilized? So do you think that could there be a play where there is an agent which just does that kind of a work where they okay they check this out, they check this out and then they bring those brands which most of the people are wearing during those hikes. Yeah, I think very much possible. This so this then boils down to the skill, right? The skill to research about a product, right? If the skill specifies that you have to look at let's say Instagram and look for not just reviews but let's say you know actual inclusion in of these shoe of the shoes in hiking photos then definitely it will do that otherwise you know most typically the skill would require them to look for proper reviews and go for the best

32:26Fit right

32:29So perfect u I have a few other things that I wanted to bring about in the discussion. So first is I think u Microsoft in their very recent build conference they announced something called as web IQ right. So web IQ is u is their AI first web search stack right and this is particularly important for developers right where as we spoke earlier right that you know search is such an important primitive for agentic work like any consequential work has to be grounded in real information right so what web IQ solves for is it sort of takes away a lot of let's say hassle in terms of making the content let's say agent friendly because web as we know is a very visual medium. It has a lot of hypertext right. It has a lot of let's say markup which the agent necessarily does not care about right what it cares about is information which is in plain text and you know in in the most unambiguous and and most consumable format right so the web IQ encapsulates all of it so it basically discovers ranks extracts and packs the information from the websites ready for the agent so for a developer you know

33:46It's a it's a 2.5 times faster So as compared to let's say a typical web search tool, right? It would take x amount of seconds. So it is now 2.5 times faster to get the right grounding information from web IQ. And web IQ still uses Bing search index but it doesn't give it it's not the Bing search API that the developer is working with. So web IQ still let's say hits the Bing search index. So it is important for businesses to make sure that their Bing search indexing is also happening. Bing crawler is is allowed on their websites and most importantly chat GPT has already started to use web IQ.

34:30So l you were saying that this would be relevant for anyone who's building an agent today. instead of solving the search themselves or building a way to you know really understand information on the website they can just directly plug it into web IQ and thereby get that whole already pre-index or or the work already done by web IQ and just plug into that for their search is that what you're saying

34:55Yeah so think of it as a more advanced web search tool

34:59Where where the output is something that the agent can directly put in context they it is not required to strip away extra information or extra extra tokens. So the it's a very token efficient output

35:13Which the developer can directly put into the agent context.

35:17Interesting.

35:18Yeah.

35:19Interesting. Okay.

35:20That's interesting. Yeah. Good.

35:21And for that again just just one last point on that. So for again just coming it coming from a brand owner or someone who owns the website or who is who's having their brand like does it have any implication on them of course just complying with what Bing Bing's API is that's one part or or how Bing search works solving for that is one part but anything additionally you think the businesses or the websites have to do for web IQ just to be better at web IQ

35:50Yeah and so if you look at the emphasis that they're putting on It's it's on the speed right. So obviously the website performance has to be good

35:59And also the website content itself has to be token efficient right so publishers should not be too verbose should also be concise and compact in terms of let's say communicating the most important information and as much as possible let's say discourage too much of let's say you know a flowery markup.

36:20Clear clear. One example alith and then again on this part was that somewhere I heard either it was something from airpulse or some or that Nike apparently used to use all their product pages. So Nike has of course a lot of products and product pages on their website. What they did was for this for the crawler and then for web IQ example it would they had created a separate page for each product page which was then given to the AI crawler. So instead of giving their regular product page they had created a separate product page as soon as they saw it's a LLM chloral which is coming to their website they just created a different page altogether. Do you think that's a good strategy in general for businesses to adopt for for websites with large number of product pages to adopt to be effic like just where you talked about token efficiency or helping LMS to consume information quickly?

37:20Yeah, in general I would say it's not a bad tactic. but

37:25You should not totally rely on u this bifurcation of traffic because you never know right the agent code can change right tomorrow GP might just have another let's say agent name and then what you're doing is basically you're serving a very unoptimized page to that agent right so these names as we have seen in you know during the course of building air we have seen you know like newer and newer agents that that we see in our data, right? So, in general, if things were static, this practice would have been good. U but it adds to the latency because there is some check that happens with whether to you know serve this page or that page that sort of takes away takes away the efficiency of the page speed. So as much as possible if you can stick to let's say you know a in the page content which is primarily optimized for the the agent we will be in the better territory.

38:27Makes sense. I wanted to highlight one one survey result right so recently there was a survey conducted of CMOs and and this was this was around their ROI from their AEO/GO efforts and their let's say you know their intent of increasing the budget or intent of let's say doubling down. So it turns out that 90 let me just pull up the real figure you know just one sec. So basically close to 97% of CMOs reported that they got a positive impact of their AEO investments and 94% they explicitly said that they will increase investments and they are now willing to allocate up to 12% of their digital marketing budgets into answer optimization. Wow, that's a huge tailwind for this industry, right? That means that there is there are businesses, there are CMOs, everyone now thinking of this as a primary course driver.

39:41Absolutely. I think the the rate at which the user behavior is changing right and are increasing dependence on u let's say AI AI engines and AI apps, right? So if you look at lot of knowledge workers are now spending a lot more time on apps like cloud co-work or codeex right and much of the discovery of new services new products even data analysis right report generation all of it happens through these apps right so nobody's actually going to even charge you for searching it is the search is happening through an agent which is part of your desktop app or or let's say it could be your personal let's say open claw So that rate of adoption has actually surprised everyone right. So there is there was this very latent and very high demand for let's say you know getting a lot of work done which was cognitively taxing but the moment these agents crossed the threshold of coherent consistency and let's say doing you know long range task I think a lot of work drastically got offloaded to them and that is making it extremely serious that you know brands and businesses

40:57Should look at their in their website and their whole discoverability or visibility for AI agents.

41:04Absolutely. And I think one more trigger to that abs with with what you alled one more trigger is people are actually seeing like most of the brands. So I was at a conference last month and 90% of the conference attendees saw a drop in their organic traffic. So the drop is real and because organic was the place where most of the last 15 years of SEO of people optimizing their website of driving traffic now as people are seeing that that castle crumble they are now trying to build okay what is then the next thing and what's coming up so I think that's another thing which is definitely driving adoption and people are think like this is one of the key key pieces for every business to solve eventually Absolutely funny.

41:52Yeah.

41:54And I think one very decent thing just yesterday like less than 24 hours ago Google announced something called as Google search profiles. Now this is this is basically Google addressing the the I'll the growing universe of or or rather exploding universe of content on social media right because there are we live in a world where social media influencers play a very important role in educating users like a lot of our information comes from let's say you know some influencer talking about something right and all of that is not making it to let's say Google search right or maybe even if it is making it's not making making it to the Google search index that easily because these are behind let's say other world gardens like let's say a meta ecosystem or X ecosystem right so Google search profile is is a way for let's say social media brands or influencers to build up a presence for Google and then you you know basically link up much of their posts on you know on other social media platforms and sort of make it available for the Google search index. So this I think is a is an extremely

43:11Power move by Google and what it also points towards is that you know a lot of social content is going to start influencing search

43:20Because quite frankly I think publishing has shifted to conversations right and conversations happen happen over video conversations or influencers talking to their audience right and it's only those conversations are getting then transcribed and making it to the to the written word right so it's I think it's a great step for Google to sort of make some accommodation for social media brands and you know getting an entry into the that that universe of content.

43:49Fantastic. And I think that's a great segue for maybe our next episode. I have very interesting three four anecdotes around that. having just spoken to a few creators recently very interesting anecdotes and very diverse views there because creators are feeling sometimes left out sometimes included in this whole agentic era and how what's what's their point of view and with Google's policies also not Google but I would say YouTube's thoughts also changing around that so I think yeah that would be a great segue for something that we discuss next time I would say

44:23Absolutely

44:25Perfect this was a Great conversation and I hope everyone listening in was able to get something valuable out of this. Definitely agentic commerce agents are going to be a big consumer of internet. So optimizing your website, optimizing your business for that layer of for that layer is going to be a critical factor in future too. Yep.

44:49Thanks Ash.

44:50Thank you Lith. Next time bye.

44:55All right. I think let me