Welcome to the MadTech podcast from ExchangeWire, bringing you latest news and views from the world of

ad tech, media, and marketing.

Hello and welcome to this very special edition of the ExchangeWire MadTech podcast.

I'm your host, John Still, head of content at ExchangeWire, and I'm thrilled to be joined today by Roman and Olena from Attekmi.

Roman, Olena, welcome from your lovely studio.

Thank you.

Thank you for having us.

Thank you, John.

Nice to meet you.

Nice to meet you too.

So before we get into the stories we're going to cover today, can you both give a brief introduction to yourselves and what you do at Attekmi?

Yeah, basically I'm Roman, CEO of the company.

And yeah, basically I'm running this company with Olena and kinda build strategy, manage everything and you know, do my best to make this business successful, obviously.

Excellent.

And you, Olena?

I am a chief product officer of the company.

We started with Roman seven years ago and it was a long journey for us from the early beginning and right now we are actually managing

everything and trying to do our best to make our clients satisfied with our product.

So yeah, that's it.

Excellent.

Well, I'm looking forward to chatting to you both.

We're going to talk a little bit more about the company later on in the show.

But first, we have two new stories to dive straight into.

So I'll start with the first story, which is that New Ozone data has shown that US and UK publisher ad request volumes fell between 32 and 41% year on year in Q2, 2026,

as AI-driven zero-click search cut referral traffic.

Spend held at better than supply, thanks to rising eCPMs suggesting that a smaller, more yield-led open web is emerging.

Yeah, these are some quite big statistics.

40% is quite a drop.

Roman, I'll throw it to you first.

How do you feel publishers can address this?

How are you seeing the shift in the open web?

Yeah, basically I'll I'll just divide my answer into two parts.

Basically, first part is that AI cuts referral traffic, obviously.

and you can build like a sustainable, successful business on the channel that you can't like don't control.

So basically I think publishers should be, should invest in direct relationships with the audience.

Basically, this must be the best,

the most important thing for them because conditions could change, platforms could change, a lot of things could change, but your loyal audience will definitely keep

your business and you safe, basically.

So I think direct relationships and transparency basically would be would be things that will, you know, be the most important.

Definitely.

Olena, how do you see this?

How publishers can really thrive in this difficult, changing environment for the open web in particular?

I would say that this uh not first time the challenge for publishers like uh we know all this information about like all the stories about cookies and you know uh then mobile uh

industry became more evolving than mob web actually traffic.

So this is not the first challenge for the publishers and they will somehow adapt and there are a lot of ways have to do this.

In terms of as Roman said, building direct relationship with your visitors, it's like the one the most key,

but also there are other possibilities in terms of technologies. Publishers can investigate new monetization channels,

let's say header bidding. More high-quality video will

actually bring the value for them as well.

And also private deals, direct deals with the advertisers, this can be like alternative how they can adapt to the decreasing the volumes.

And so as from from my perspective, maybe it's not about the volumes anymore.

It's about the value of the content and the value of the publisher.

Yeah, yeah.

And you know, the more the more direct is relationship, the the better would be quality and one impression will bring more revenue, basically.

And if we are talking about programmatic, so we, all of us know about supply pass optimization, so this is can be also the key point for them to try to decrease the

hopes and to try to reach out to the advertiser more directly and yeah it can bring the value.

Yeah, yeah.

So no one wants to pay for intermediaries.

They still looking for a direct you know.

Yeah, I think there's a couple of things that you've mentioned there that are really important.

One for me is, from I've heard from a lot of publishers, is the change from volume to quality.

It's now, you know, it's about how invested your audience are, how much time they're spending with your content rather than just sheer volume of clicks.

Are you seeing that shift in your work with Attekmi?

How is this altering what you're doing?

In terms of development, you know, we are very open to any kind of suggestions from publishers.

If they have some pains, if they have some needs, this is something that we can develop for them and try to build the technology that will help help them monetize efficiently.

So obviously then maybe in this case, the context, the taxonomies, everything that like

about the enrichment and how to like bring the value to those less volumes of requests that they are running at the moment will be very important.

So in terms of settings, in terms of filters, in terms of post-processing, that's something that we can develop for clients who are searching for that.

Yeah, yeah.

So the most important one of the most important things in this whole discussion and the like overall is interoperability.

And this is really important for us.

And basically the product development at Attekmi is built basically like one of the key pillars is feedback of our clients, partners.

So basically we don't do anything without the approval of our, you know, partners.

We consider our clients as partners because everything that we do is focused on their

revenue grows and we want to provide them with a solution that allows them to be transparent, to manage SPO properly, and yeah so this is something we do.

We want to provide value if we want our partners to succeed because if they make a lot of money, you know, we make a lot of money.

So this is how it works basically.

That's a happy relationship, yeah.

Win-win configuration of this.

Yeah.

Exactly.

In terms of how your company's developed and how you've seen things adapt,

where do you stand on the future of the open web and the future of our programmatic age in this new agentic era?

How do you think it's going to play out?

Well, this is this is super exciting to be honest, because programmatic and AI and LLMs and everything is like something that is working together.

So this is not like programmatic will disappear or die.

It's not like AI will just kill everything that we know, like open web and programmatic.

I think this is super cool that publishers and brands and other people use AI a lot because this is just, you know,

comfy and it may it takes less time like to find something, I don't know, to compare goods, to pick the best platform for you in the marke t, for example, so you just use AI.

But this is just, it's not like another like channel, another way. It's just an additional thing to programmatic, additional thing to open web because AI needs to learn,

AI needs content that is, you know, open web.

So it's just something that help each other to make a really cool, unique, complete advanced ecosystem, in my opinion.

Yeah, so you see it more as a, it's more of a tool to help programmatic rather than a competition or a threat to programmatic.

It's not a threat, it's just adjacent tool that, you know, helps.

Maybe this is the, let's say, opportunity as well.

because I do believe that there are still open web publishers which bring the value to the visitors and, let's say, if we are talking about some news, about some expert,

experts in any sphere, so there are should be like those top quality publishers that will still have visitors.

So actually just like

the maybe key challenge here is to forward this value to the advertiser and maybe for this they will need more information, more enrichment more context

fields that we are using in the programmatic maybe if will they like show that this placement, that this inventory is very valuable for them. In this case, it will be

it will actually have success on the demand side and on the advertisers side.

Excellent.

So yeah, more of an opportunity than a challenge.

I like that framing of it.

Yeah, yeah, it's you know, it's an opportunity and at the same time it's a little bit of challenge, but it's like because you you you have to you have to structure everything, you

have to unify everything.

It's like in programming in RTB for example, everything is kinda regulated more or less.

Yeah.

But this new thing, this new thing with LLMs, with AI, it's kinda really new.

So we need to play around with it and configure, you know, how to

how to work with this properly to get more value.

Yeah.

It's, think, like any new technology, it's a case of working out how best it works for you and what benefits it can really bring.

And I think when we see numbers like this in terms of traffic heading to sites, it is difficult to frame it with an upside.

But like you say, there are the focus on quality, the changing metrics.

It does offer a lot to the industry as well.

So yeah, an interesting challenge slash opportunity for the industry to face.

Yeah.

Yeah, obviously.

Excellent.

Well, I'll move us on to the next story.

And speaking of LLMs, it looks like OpenAI's ad business is on pace to miss its own forecasts for success by about 90%, which is a fairly huge miss.

Emarketer has projected that standalone chatbot ad revenue across ChatGPT, Copilot, Google, Alexa, et cetera, will hit just $5.41 billion by 2030.

OpenAI's own project,

was meant to be hitting 100 billion in a target.

So did OpenAI overestimate how much?

It seems so.

Olena, what do you think?

Were they a bit too bold with their forecasting?

Yeah, just like the this is very very, sounds very attractive, let's say, for investors, but if you are talking about the numbers, we are not sure because we cannot

actually guarantee how users will still use those conversational and AI, if they will see start if they will start to see ads.

So will it be

good for users to pay extra money, for example, to subscription without ads?

Or can we be sure that there will be no any hidden advertising in those answers from, let's say, ChatGPT?

So we don't know for sure. Yeah like, generally it's also opportunity, like opportunity every new ad format and

AI advertising this for sure should be as a new ad format, but we don't know exactly how advertisers will evaluate the value of this new ad format.

How the publishers can like pay for that to be promoted.

Will we see this promoted ad answers or not promoted?

This is influenced so much.

And for me personally, I

I am not sure that I will prefer to use advertiser answer.

I will prefer to make my own research in this case maybe.

So I'm just trying to like think at the user in this case.

So I don't think that those numbers are that promising as they are trying to sell.

Yeah, think you've hit on a few points there that are so important and so interesting.

We had a lot of discussion earlier in the year about how it would work, what the formats would be like, and that kind of idea of whether the advertising side of it really impacts

the answer you get and whether it's trustworthy, whether it's going to be labelled.

There's so many questions for the ad industry.

Roman, how do you feel about OpenAI's forecasts?

Were they a bit over the top?

Well, I'd say they are bold, obviously, and they are positive, you know.

Because when you build startup you wanna you wanna show, you know, cool numbers, a lot of opportunities,

and you wanna attract more investments, more angels, more blah blah blah.

But I think like there are two sides of the coin basically.

So the first side, on one hand, AI is super precise and

everything will depend on the behavior of the users.

So for example, if the if the user is looking for something like, I don't know, the best CRM platform for the SMB and during the research with the AI, for example, Claude

or I don't know, ChatGPT, ChatGPT natively suggests like the best option, that probably that would make sense.

I don't know.

But I think that must might be a good working model, let's say.

But on the other hand, if AI shares with you a lot of redundant things and unnecessary things, but just with one aim to sell you something that you don't need, then

probably that won't work.

So I think that the execution is the key and the most important thing is result, is

if everything works correctly and the user sees value, then that would be a good nice thing.

But as I said previously, I think that AI would be just a companion to the open web and to the programmatic ecosystem.

So they will build together one like piece, uh one really good solution for for buyers and sellers.

Yeah.

How have LLMs fitted into your life?

How do you use them?

And where could advertising bring benefit to that

do you think?

Are you using them frequently?

I use it every day, like it's on a daily basis, but basically I use AI just for optimization,

getting rid of different redundant things that are repetitive in my day-to-day work.

At this very moment I'm not using AI for I don't buying stuff or something like that.

I prefer, I'm old school, I prefer to do a research on my own.

But yes, AI is is involved in my life, like personal life.

And on top of that, obviously we work on the AI tool uh at Attekmi.

And this year we'll present our new feature, a new tool, AI tool that helps our partners.

This is like advisor.

Advisor and data analyst in your pocket.

So we wanna utilize AI as a key tool that helps to improve and optimize activities.

Olena, do you use LLMs in your daily life?

Are you a daily user?

Yeah, I'm also daily user, but I am using ChatGPT, for example, for small researches.

In terms of work, it's different kind of approach because our actually we do use AI to optimize manual routine tasks.

But what is important even if you are talking about working with the data and if you are

willing to receive some insight from the AI, it should be moderated by a human because yeah, AI can advise you how you can change the settings, how you can change the

volumes, what you should take a look at the moment, but the decision should be made by actually user of the platform.

So this is, and we are moderated this on our end because the

traffic, live traffic is very sensitive, the settings are very sensitive, so AI is just a helper, not a key value, not a key player,

but a helper that will advise you how to optimize.

But you are the main on your platform.

Yeah.

So basically we're technology driven but human operated.

I like that a lot.

I think that's very important that there's always the element of human oversight.

I think AI can be trusted to do a lot of things and can be a very helpful tool, but it's always going to need human decisioning and human oversight.

That's good to hear.

So the LLM market, these figures are showing that maybe they overestimated.

Like you said, Roman, it looks good for potential investors and shareholders to overestimate, but...

there's still a market for it.

5.41 billion is still a large amount by 2030.

Olena, how do you see the LLM market progressing?

Do you think there's new formats, innovations?

How do you think it will develop over the next few months and years?

It's as I said actually hard to predict how it will be working because we don't see any like protocol.

So I used to work with protocols.

I would like to see a protocol how it will be working, then I will be able to tell you for sure how it can be adapted to,

how it can be implemented, and how this monetization will work for sure.

So yeah, let's wait.

Let's wait maybe once we will have some

kind of... so we are seeing right now some agentic AI agents that some publishers and advertisers are trying to develop on their end and test but we don't

see the value at the moment.

So everyone is just talking, would like to test but when we like would like to test with something specific, with someone specific, they don't have idea how to manage this.

So this is just

an early stage of this programmatic and not just programmatic AI monetisation sphere.

So I believe that we will need some time to see how it.

Roman, do you feel,

How do you feel?

Well, I feel I think like almost the same way because obviously new formats will be live.

I don't know whether it it will be soon or not.

I think like a couple of months will show us because I think that new formats will definitely appear in ChatGPT and Claude

because AI will definitely recommend you something.

They need to show something to you, so you have to see this.

But everything should be more native, not just like, you know.

pops up out of the blue and this is just distracting.

So new formats will definitely like come up

but hard to predict how everything will look.

I think something hybrid will be at place, but we'll see.

I think a couple of months will show us new formats, near purchase.

Yeah, it's hard to predict when everything is changing so rapidly, so popped you both on the spot a little bit with the prediction there.

But I think it's very much a wait and see approach on what formats really develop through LLM advertising.

But it's certainly something that's with us.

Publishers and advertisers should just play around with those formats and those new approaches because I think that this is something that will be visible in programmatic

ecosystem like real quick because there are new experiments, we see that and we know how to work with that, but I think like this is not something that is like stable now.

So just we need a little bit more time to see new, no common formats.

Yeah, and it's that test and learn period.

I think there's a mixture of excitement and fear probably for everyone.

And the test and learn period is essential to make sure the advertising industry learns how to use this properly.

And standards are probably necessary, but we'll see how it develops.

Excellent.

Guys, thank you for your input on both those stories.

So interesting to get your takes.

Before I let you go, I want to talk a little bit about Attekmi.

So tell me.

First off about the guys, about the company and what are the big issues that you're solving for and also that you're seeing in the market today?

How is everything at Attekmi?

Well, you know, it's hard to to tell you everything like in a couple of minutes because this is like lifelong story

because basically we started with Olena eight years ago

but the company itself exists like almost ten.

Anyways, basically we build a lot of things and now we manage a lot of things and now we are dealing with a lot of problems, but I would rather tell, it's not

problems, it's challenges because because yeah.

Well the tech means it's we say it's your partner, not just the vendor that provides you with some solutions like some platform, something like that.

We are ad tech partner

and we can solve pretty much all the issues, all the challenges, all the ideas that you have in regards to the ad tech development, because we can develop basically like

everything.

SSPs, SSPs and ad exchanges and DSPs and like whatever you want.

If you wanna build some rocket ship in the ad tech industry, we'll definitely build you the rocket ship.

We have our own products, we have white label solutions, we can build everything from scratch if you need, and obviously if you have enough money.

So basically, we have vast experience in ad tech and we can utilize it and we can make it work, and we can make it,

we can bring value to the table immediately.

And if you want to make money, we can help you to do that.

Because your success is our success.

We consider all the clients, not just clients or users or I don't know, just we call them partners because we partner up, we are involved in activities, we align our

strategies with our partners' strategies, and we do everything like to make your ad tech journey successful.

So this is like in a notch so.

Regarding challenges, you know, we have a lot of challenges.

We had a lot of challenges, but we successfully dealt with them

and as we discussed today, so AI and LLMs basically are like at the same time challenge and opportunity.

So we implement AI to the products, we implement AI to the new custom platforms, new technologies that we build.

But at the same time, we just missing little piece of the puzzle, which is you know, regulations.

So

once regulations go live, it will make a little bit easier our lives because now we do a lot of researches, we experiment a lot, we talk to partners, we talk to vendors, we talk

to the industry, and basically basically yeah, this is pretty much it.

Because I'm like loading this, you know, propeller.

So I'm just gonna stop and and let Olena talk.

No, I thought that was very clear.

Building rocket ships for the ad tech industry.

Olena, how are you seeing the industry?

How has it changed since you began with Attekmi?

I I do remember those times when uh at Ads.txt and Fly Chain were like a horrible story for everyone.

Then it was like cookie and cookieless solution then rapidly evolving after Google decided to change everything.

So we do see changes in this industry and they are like every year, every couple of months as we trying to adapt and trying to develop new policies, frameworks,

regulations and everyone on the market is trying to do the same.

So and we are helping our partners to manage everything because we do have the expertise in this sphere.

But I would also add to those things that Roman said.

Everyone is talking right now about those agentic AI.

And we were at Cannes Lions this summer and everyone were asking what are you doing, what are you going to do with that?

And yeah there are still a lot of questions like publishers I believe they do do want to understand what decision will be made on the

advertiser side, if there will be agent who is choosing which inventory is the best, so okay, what will be the key metrics to calculate them on the demand side and they want

to have all the system information and if we are talking about the advertiser they are not sure that let's say a publishers agent will be able to fully replace

Let's say our programmatic ecosystem when we have a lot of fields, a lot of information, those cookies and cookies, everything.

So we don't have these answers yet.

We just have, as Roman said, opportunity and challenge at the moment.

So let's let's wait.

We already explore a DCP protocol, but it's still under discussion, not implemented yet on most companies' side.

So

this time we need to see.

Yeah, it's just you can't find a lot of agents built by a lot of companies, just a few of them.

And you can check them out on the website because they are visible, you can just, you know, check out and see that this is just a few of them.

So basically not a lot of companies know how to how to build it, how to implement it.

And I'm not gonna lie, I don't know either, but we we are working on that.

We are testing, we we are talking with our partners, our clients, our vendors, like

And yeah, so something is cooking, but I'm not gonna tell you what.

Fair enough.

I would like to know what and I'm looking forward to talking to you again down the line when the cooking is done.

Thank you both so much for your time today.

I really appreciate you joining me.

It was really great to get your insights and I'm very jealous of your studio.

So thank you very much.

Thank you for having us.

Was for nice talking to you today.

Thank you for great questions.

Have a nice day.

Thank you again and thank you all very much for listening.

That was this edition of The MadTech Podcast.

You've been listening to The MadTech Podcast brought to you by ExchangeWire.