Agentic Advertising With Ben Williams on The Six Sells Podcast

What is agentic advertising, how does it work, and what could it mean for the future of media buying? Ben Williams, Founder and MD of Kiln, joins Mike Nicholson and Fern Potter on The Six Sells Podcast to discuss buyer and seller agents, the infrastructure being built around agentic advertising, transparency, publishers, testing, governance and how the advertising Lumascape could change.

Episode summary

In this episode of The Six Sells Podcast, Mike Nicholson and Fern Potter speak to Ben Williams about agentic advertising and the potential impact of AI agents on how media is bought and sold.

Ben is the Founder and MD of Kiln, a technology agency for the media industry. Having started Kiln in 2019 as a technical business development agency, Ben and his team increasingly moved into product, engineering and transformation work. More recently, that has included agentic AI briefs and helping media businesses understand how emerging agentic infrastructure could fit into their organisations.

The conversation explores what agentic advertising actually means, how much of it is real today, and why Ben describes 2026 as a year of experimentation. They discuss buyer and seller agents, storefronts, AdCP, the AAO, Prebid, Scope3 Interchange and how publishers can begin testing agentic transactions without assuming the whole advertising ecosystem will change overnight.

The discussion also considers some of the problems agentic advertising could potentially address, including transparency of spend, programmatic take rates and the friction between buyers and sellers. At the same time, Mike, Fern and Ben discuss the risks of repeating some of programmatic advertising's mistakes, including bad actors, publisher economics and an industry that can sometimes optimise problems rather than solve them.

They also look further ahead at the future of holding companies and independent agencies, AI-native businesses, new media environments, changes to the Lumascape and Ben's belief that changing browsing behaviour, hardware and software could ultimately make agentic advertising much more significant.

Topics discussed include:

• Ben Williams' career and the evolution of Kiln from a technical business development agency into a technology agency for the media industry
• Kiln's approach to agentic AI projects and briefs
• What agentic advertising means and how buyer and seller agents could interact
• Why Ben describes 2026 as a year of experimentation
• AdCP, the AAO and emerging agentic advertising standards
• How agents could help publishers discover incremental demand
• Better briefing, discovery and relevance between buyers and sellers
• Transparency of advertising spend and the economics of the open internet
• The concept of "human + tech = the deal"
• Brand agents, seller agents and storefronts
• Human intervention and oversight in agentic media buying
• Scope3 Interchange, Prebid and different routes into agentic infrastructure
• Why publishers are not necessarily missing meaningful agentic revenue today
• Prototyping, revenue tests and small experimental budgets
• The role of SSPs and other adtech businesses in building agentic infrastructure
• Testing and learning before agentic advertising reaches scale
• The potential for higher CPMs and a greater share of spend to reach publishers
• Holding companies, independent agencies and changing approaches to experimentation
• Why commercial rigidity can make innovation more difficult
• New media supply and audiences, including creators and sports
• The transition from analogue to digital and from digital to AI
• Trust, governance and confidence in emerging agentic systems
• AI-native businesses versus businesses treating AI as a tactical add-on
• The importance of data foundations before building AI products
• Collaboration between independent UK technology businesses
• Which brands are currently driving experimentation in agentic advertising
• Fraud, bad actors and governance in an agentic ecosystem
• How agentic advertising could change jobs rather than simply remove them
• Brand, measurement, signals, creative and seller agents
• How agentic advertising could change the advertising Lumascape
• Why existing advertising infrastructure is likely to evolve rather than disappear immediately
• Ben's view that browsing behaviour, devices and the internet itself will change
• How agentic systems could make advertising more accessible to smaller businesses
• Why brands may increasingly try to become part of the content and answers surfaced by AI systems

What is agentic advertising?

Agentic advertising is an emerging approach to media buying and selling in which AI agents can perform parts of the workflow currently handled by people and traditional advertising technology.

Ben Williams describes an ecosystem in which a brand agent can contain a brief, seller agents can make media products and pricing discoverable, and agents can potentially negotiate with one another before a human confirms or changes the resulting plan.

The aim is not simply to automate today's programmatic advertising workflow. Ben sees an opportunity for agents to help buyers discover relevant media opportunities and for publishers to receive incremental demand that they might not otherwise have found.

He also expects humans to remain involved. The systems being tested today still include significant human intervention, and Ben describes the emerging model as "human + tech = the deal".

Is agentic advertising actually happening in 2026?

Agentic advertising is happening in 2026, but Ben Williams describes the market as being in an experimentation phase rather than operating at significant reported scale.

Ben says publishers and technology companies are prototyping agentic systems and beginning to run live revenue tests. He compares the current stage with the early days of mobile advertising, when brands would allocate relatively small test budgets to understand a new channel before committing significant investment.

He does not believe publishers are currently missing substantial campaigns simply because they do not have a seller agent. Instead, he argues that businesses should use this period to understand the technology, prototype it and test real transactions before the market develops further.

Ben says there is infrastructure emerging around the opportunity, but there is not yet much publicly reported data showing large advertising budgets flowing through agentic systems.

How could buyer and seller agents change media buying?

Buyer and seller agents could make media discovery, briefing, negotiation and transactions more direct between advertisers and media owners.

In the workflow discussed by Ben Williams, a brand agent can contain information about the advertiser's brief and requirements. Seller agents can then expose relevant products, inventory and prices. Agents can potentially negotiate before humans decide what should ultimately be bought.

Mike Nicholson describes a potential future in which a brand explains its challenge once and a system identifies relevant sellers based on that brief, rather than requiring the brand to repeatedly explain the same problem in meetings with different vendors.

Fern Potter argues that the same principle could extend further upstream. Better AI-supported briefing could improve the information that reaches the agents in the first place, increasing relevance before the transaction begins.

What are seller agents and storefronts in agentic advertising?

A seller agent represents media products and opportunities so that buyer agents can discover and potentially transact with them, while a storefront provides a structured representation of what a media owner has available.

Ben Williams describes a storefront as effectively taking information that might traditionally sit in a publisher's media kit and data solutions and making it accessible in natural language to agentic systems.

Publishers can approach this in different ways. Ben discusses custom seller agents built around Prebid as well as more off-the-shelf approaches such as Scope3 Interchange, which can help businesses connect agentic buying and selling systems with existing advertising infrastructure.

For Ben, getting a seller agent in place is partly about ensuring a publisher can participate as agentic demand develops. He compares the potential incremental demand with the unexpected incoming advertising bookings that sales teams once received by fax.

Could agentic advertising improve transparency for publishers and advertisers?

Ben Williams believes greater transparency of advertising spend could become one of the important benefits of agentic advertising.

The discussion reflects on longstanding concerns about how much advertiser spend ultimately reaches publishers through programmatic supply chains. Ben argues that more direct interaction between buying and selling agents could reduce some of the opacity and friction between the two sides.

The theory discussed in the episode is that publishers could potentially receive both a greater share of the CPM and, in some cases, higher CPMs because agentic transactions could behave more like direct buying.

However, the conversation also acknowledges that these outcomes are not guaranteed. Mike raises the concern that publishers could have less ability to challenge buying decisions if machines increasingly decide where money is spent. Ben, Mike and Fern agree that transparency, governance and human oversight will therefore matter as the market develops.

Will agentic advertising replace people working in media and advertising?

Ben Williams does not expect agentic advertising simply to remove people from media and advertising, but he does expect roles and skills to change.

He sees agents as a way of making existing teams more powerful and expects new jobs to emerge around building, operating and negotiating through agentic systems.

Ben illustrates the change by comparing his own first media sales job with that of a younger colleague. Ben started by making around 60 sales calls a day to hotels. His colleague Giovanni's first entry point into media has involved building seller agents for clients.

The tasks are different, but people remain involved. Ben expects workers to have to upskill as the technology develops, just as they have during previous shifts in the media industry.

What could agentic advertising mean for publishers?

Agentic advertising could give publishers access to incremental demand and more direct relationships with buyers, but the eventual impact on publisher economics is still uncertain.

Ben sees seller agents as a way for publishers to make themselves discoverable to buyer agents and potentially receive business they would not have found through their existing sales operation.

Fern Potter hopes this could improve the flow of advertising money towards the businesses creating the environments in which consumers actually experience brands.

Mike Nicholson raises a counterpoint. In traditional media sales, publishers could speak directly to an agency when they believed they had been excluded from a media plan. If automated systems increasingly determine buying decisions, publishers may need new ways to understand why they are receiving less spend and how to challenge those decisions.

The episode therefore presents agentic advertising as an opportunity for publishers, but not an automatic solution to the economic problems created by previous generations of advertising technology.

How could agentic advertising change the advertising Lumascape?

Agentic advertising is likely to change the functions and boundaries of companies in the Lumascape rather than immediately remove the entire existing adtech infrastructure.

Ben Williams says the agents currently being developed include brand agents, measurement agents, signals agents, creative agents and seller agents, with reporting also forming part of the workflow.

He expects these systems initially to operate on top of infrastructure that the advertising industry has spent roughly two decades building. He compares it with broadband infrastructure: new capabilities use the existing foundations while those foundations are gradually upgraded.

As more money moves through agentic systems, advertisers may begin asking harder questions about what existing technologies such as DMPs, CDPs, DSPs and SSPs are actually doing and which functions remain necessary.

Ben therefore expects both contraction and expansion. Some categories and companies may consolidate, while new businesses and new forms of technology emerge.

Why are independent agencies and AI-native businesses important to agentic advertising?

Independent agencies and AI-native businesses can be important to agentic advertising because their structures may allow them to test new technology more quickly.

The conversation contrasts the agility of some independent agencies with the commercial rigidity that can exist inside large holding companies, where established trading arrangements, technology choices and organisational silos can make rapid experimentation difficult.

Ben also sees opportunities emerging outside traditional digital publishing. Creators, sports businesses and other organisations that are not constrained by legacy programmatic infrastructure may be more willing to build for new forms of media consumption.

The same distinction appears in AI transformation more broadly. Ben says businesses sometimes want an AI product without first addressing the data foundations required to make it useful. AI-native organisations, by contrast, can build their data, software and workflows around AI from the beginning.

What are the risks of agentic advertising?

Agentic advertising could reproduce problems including fraud, hidden take rates and poor incentives if the industry does not build effective governance into the new ecosystem.

Ben Williams expects bad actors to appear wherever significant money flows. That could eventually include rogue seller agents or businesses finding new ways to extract excessive fees from transactions.

At the same time, he says much of the current development work is collaborative. Engineering teams from across the industry are working together on standards and governance, with an emphasis on protecting the health of the emerging ecosystem.

Fern Potter argues that the experience of programmatic advertising should act as a cautionary tale. If agentic advertising develops at scale, the industry has an opportunity to learn from previous problems rather than simply rebuilding them in a new technological form.

What is Ben Williams' big bet on the future of agentic advertising?

Ben Williams' big bet is that browsing behaviour and the user experience of the internet will change, making agentic advertising more logical within a new technological paradigm.

He expects hardware, software, browsers and infrastructure to evolve together. Rather than viewing agentic advertising as an isolated advertising technology, Ben sees it as one part of a broader transition from the digital era into an AI era.

He believes that by the end of the decade smartphones and browsing could look significantly different from today. If that happens, agentic systems could become a more natural way to connect users, brands, media and commerce.

Ben also sees two potential economic opportunities. The first is giving brands greater confidence in where and how they invest advertising money. The second is making sophisticated digital advertising more accessible to small and medium-sized businesses.

What is The Six Sells Podcast?

The Six Sells Podcast is for people working in marketing, media, advertising, adtech and technology. There are over 150 episodes (and counting), and the biggest episodes have achieved 30,000 views across all platforms. Want to get in touch? Email hello@sixsells.co.uk

Who is Six Sells?

Six Sells is a UK-based executive visibility and thought leadership agency that works globally with CEOs and senior leaders in media, advertising, adtech and technology.

Founded in 2018, we turn each executive's expertise into distinctive, authoritative content in their own voice, built to reach two audiences at once: the humans that matter to them, and the AI models those humans now ask for answers.

Our work combines executive ghostwriting, video, podcasts, photography and awards to build familiarity, trust and authority across LinkedIn, Reddit, YouTube, Substack, the trade press and the wider web.

We also help executives and their companies improve their visibility in AI search and large language models (LLMs) through generative engine optimisation (GEO), increasing the likelihood that platforms including ChatGPT, Google Gemini, Perplexity and Claude can discover, understand, cite and recommend their expertise.

Read more here, or email hello@sixsells.co.uk for more information.

Full episode transcript:

Mike Nicholson: Ben, Fern, how are we doing?

Ben Williams: Very well. Good, thank you. Thanks for having me.

Fern Potter: Good, thanks for being here. Nice to see you again.

Mike Nicholson: Could you give us maybe a little bit of a potted history, 30 seconds on your career to date and what you're up to now, please?

Ben Williams: Absolutely. I will go as fast as I can. I started off selling classified ads at Haymarket, got into digital at VNU, worked for an online video startup, then helped launch Project Kangaroo back in the day. Then went and helped in the movie launch across Europe. Then worked at Apple. Then worked at Channel 4 as digital director for the Box Plus Network. And then in 2019, I started my own business, which is called Kiln.

Mike Nicholson: Kiln. I love it. We don't need to hear from you by this point. I think people know who you are now.

Fern Potter: They do. Yeah.

Mike Nicholson: Less airtime. I think that was wise. Fern, confirm what we're going to talk about.

Fern Potter: Well, I am so happy to have you here because I adore everything that you do. You are the OG, I would say, of agentic advertising, and the work that you're doing at Kiln always really impresses me, the team there.

Ben Williams: Thank you.

Fern Potter: And how you're pioneering in agentic advertising. So I think the purpose of today and what we wanted to talk about was just that: what it is, what's happening. There's probably demystification required in terms of what's working, what's not.

I think, and we were chatting about this before, it's hugely transformational for brands, advertisers, consumers, and more transformational than any shift that we've seen recently. I think we've seen a lot of iterations in programmatic, we've seen a lot of iterations in social media, even actually just talking about AI and agentic, that's quite iterative. But this feels really transformational because it's actually changing the whole foundations and the way organisations buy and sell media.

And I think it's really fascinating to get your expertise, as a founding member and partner of AAO as well, as to what it is, what we're doing and how's it going? So that's kind of our chat today, if you agree.

Ben Williams: Well, I'm here. I'm here to riff with you both. So whatever you want to talk about, we can talk about. So do you want me to respond to that?

Fern Potter: Please do. Yeah. And obviously talk a bit about Kiln and the part you're playing in that as well.

Ben Williams: Thank you. It's an interesting adventure. So, what our relationship is, you've been a client of Kiln.

Fern Potter: Yes.

Ben Williams: And we've also now partnered up and we're working on a couple of things, aren't we? Obviously, we won't talk about it here because they're kind of in motion.

Fern Potter: Yeah.

Ben Williams: And so you would have seen firsthand our approach to how we look at agentic AI projects.

If I go back, in 2019 I founded Kiln. Back then it was a technical business development agency for startups. That's what it was back in 2019. And our first client was an in-game audio platform. We helped them go to market and introduced them to some big game developers and things like that.

And then what happened over time, as we started to go up through the gears of the industry and started to work with bigger customers, is that there was a real need for product and engineering expertise. And so I started working with Andrew Gusev and Dimitri Spinov. Hi, Dima. He wants me to say hi, Dima, whenever I'm on a podcast. I forgot to do it on the one we did a couple of weeks ago. So, hey, Dima.

They're great because they're based out in Slovenia and they were, I think, on the early team at AppNexus, not directly, but like an outsourced dev team. So they've been involved in the engineering of the monetisation market for a long time.

And then we've done a whole bunch of projects together. So there was a big period pre-AI where we were doing custom transformation projects for clients.

Fern Potter: I was going to say, I don't remember pre-AI anymore.

Ben Williams: I know, right? It feels strange saying that. It wasn't that long ago. Everyone's like, "Oh, can I have a product roadmap?" and ChatGPT or Claude or whoever will go, "Here's what we think you want", and then try to put that into implementation. Good luck with that.

But we're just going through that period at the moment. So more recently, what we've been looking at is, well, what does being a technology agency for the media industry, which is what we are today, mean today? And it means a whole bunch of stuff.

One of the things that we're looking at is agentic AI briefs. We got our first brief on that last June. I won't say who, but it's a big global tech business. So that's when we started to hear things about network orchestration and agentic workflow.

The reason why we get pulled into it is because we tend to get the briefs for what clients are struggling to do internally for whatever reason. So we kind of get the stuff which is like, "Oh, we're trying to launch this in six to 12 months." And then we have to help clients with that. But the team are good enough to be able to stay ahead.

Coming back to your point, agentic advertising really got going in October last year with the launch of AdCP. And then the governing body, AAO, launched soon after that, which Kiln is an active member of. I'm in a few of the working groups.

Really, what it is, is there's a belief that the next trillion dollars of advertising will be through agents, not just people. I don't think anyone's going to lose their jobs. I think people will have to upskill, of course. But that's just normal, right? That's just how industry goes. The next big growth, I think, from the industry is going to come from agentic advertising. And we can get into the detail on that as we talk.

Fern Potter: Mike, agentic advertising to you, what's your read on what it is and where it's at?

Mike Nicholson: I don't really know, to be honest. I speak to a lot of people that say that they're moving towards it and that it's a reality today. But I don't know how much of that is PowerPoint and how much of that is actually infrastructure. So it'd be interesting to get your view on that.

I know of some clients that are actively doing it, but I don't know what percentage of the overall. When you see it, are they more on the buy side or the sell side? Is it more like publishing?

Ben Williams: More on the sell side.

Mike Nicholson: So I know there are some SSPs doing it, and if there are SSPs doing it, there's obviously some DSPs doing it, but I don't know the scale of it. I think it feels like we're talking about something that's going to happen in the future, but maybe with some tests running now. Is that fair?

Ben Williams: No, I would say 2026 has been the year of experimentation. And you've got two camps. They're not competing. It's quite funny to hear about what the public perception is. They're not competing.

You've got ARTF and AMP from IAB Tech Lab, and then you've got AdCP coming from AAO. In my simplistic mind, AdCP is more about discovery of buyer agents to seller agents.

Having come up through the media sales route at publishers, you're always looking for new business, right? That incoming call: "Wow, I hadn't heard of that company that just booked a double-page spread back in the day." Well, how can agents actually find new business for publishers, and good quality, good pricing as well?

Because you see some of the CPMs that publishers get from programmatic and you're like, "What? Really?"

I was talking to Dan at Heineken in this room earlier today about this. I was saying, wouldn't it be good if there was a platform where, on the buy side, all of the brands could say, "These are my challenges, these are my briefs", and I'm just going to put it in once.

And then that interacts with all of the different sell-side vendors that are plugged in on the other side and goes, "Right, you should consider speaking to this. Here's a shortlist based on what you're trying to do versus the capabilities, the claimed capabilities, of all of the sell side. You should speak to X, Y and Z."

Mike Nicholson: Yeah, because the problem is, to get what the brand's challenges are, you have to get their attention in a meeting in the first place. There's only so many times brands want to do that and then explain again what their problem is.

But if they could do it once and that could in some way be interfaced with what can be delivered on the sales side, then that would be really interesting, I think.

Ben Williams: And I think that is the philosophy of what's happening. I think that's what there's a movement towards.

Fern, we're talking to a business in APAC, obviously who we can't name yet because it's embargoed, but that is looking at that. It's how can publishers interact better with agencies and are they set up for that?

The founder is really looking at what agentic means because he's building for the next one, two, three years. He's a very smart chief digital officer. Agentic is his bet, essentially.

Fern Potter: Yeah. I think that's quite right. You mentioned quite a few acronyms as well. We might have to unpack some of those.

Ben Williams: Sorry about that.

Fern Potter: We've got ARTF and AMP, that's the IAB Tech Lab. And then we've got the AAO, which is buyer-seller agents on the agentic side, AdCP.

Ben Williams: AdCP.

Fern Potter: That was the one I'd missed out. And that's really about that activation component. That's like the mechanics of buying and selling.

But I think, to your point as well, how do you then start taking the same principle and building it upstream to get better briefing, better relevance and get breadth easily of responses? I think that's what's exciting.

And as you said, a partnership that we've got as well is to, yes, there's the activation element of it that is foundational to the change of how we transact between the buy side and the sell side, maybe actually stop being sides to a degree. But to get to putting that brief into the hands of the agents, you've got to make sure the brief is right as well.

So how do you take some of that beautiful logic and that speed that AI gives you and the breadth of discovery and build that upstream? Which is something that IA are working on in tandem to be able to bring all that together into what I guess is the foundation of what's agentic advertising.

Hopefully what it means, and this is my personal view, is just really shit-hot brand experiences for customers online.

Ben Williams: One hundred per cent.

Fern Potter: I think we disintermediate that with tech a lot. I'm guilty of it because I worked in adtech for so long, but that's one of the reasons, kind of coming out of it, to go, okay, well, the tech's been disintermediating the experience because there's so much of it and it's technical and there's a lot of friction between the technologies.

And to your point, there's not always the right remuneration strategy in place for those where the user's actually having that experience with the brand online, on the broadcaster website, on the publisher website.

I think my hope for this transformation is that better experience and, obviously, a really positive way in which advertising dollars go to the right place based on that experience.

Ben Williams: If you think about the internet, let's just say, and don't quote me on this, roughly 60% is monetised through ads, and then the rest is split between e-commerce and subscriptions. There are different models. So the majority of the funding of the open internet is from ads.

Look at Meta's valuation, which is a great point in that, right? Google and everyone else. Apple's advertising business is going great guns now with ad platforms and whatnot. So big tech are realising that ads are a great business model, and I think it will grow.

But it's just been through a bit of a murky few years. There was that famous PwC report. It's fresh in my mind because I shared it with a CTO client recently. How much money was lost in the internet, where they measured it and put a load of budgets down and just went down the pipes and then checked on the reception.

Fern Potter: PwC, yeah.

Ben Williams: About 50% got to publishers.

Fern Potter: Yeah.

Ben Williams: And it's like, wow, hold on a minute. That's not right. And then you go and work with publishers and they're firing everyone and revenues are down and it's profit warnings, and it's like, whoa, this isn't right.

We don't want a dead internet. We don't want an internet where it's just AI slop.

The P&Gs and the Unilevers and other advertisers, if you talk to them, are always a bit reticent about programmatic because they're like, "Yeah, it's great. It's kind of promised all this lovely stuff and this targeting. But is the money actually reaching the end person that's going to go and buy my soap or beer or whatever's being advertised?"

I think one of the big use cases for AI in general is transparency.

One of my good friends, I won't name him, but he's a really top CPO at one of the top 10 publishers in the US. I remember asking him a couple of years ago, "What's your view on AI?" ChatGPT was about three or six months old. And he's like, "I imagine AI would just be a ring around the repository that is the internet. We'll be able to use it to find information much quicker."

If you then look at the use cases for agentic, I think the transparency of spend is going to be really nice. That's why I'm excited about it. It's the transparency of the spend.

Mike Nicholson: The thing that I work with, I'm similar to you, Ben. I sort of grew up in media sales and publishers, back when things were printed.

Ben Williams: Which ones, Mike?

Mike Nicholson: Telegraph, Evening Standard, Metro.

Ben Williams: Consumer publishing. I was B2B. You were consumer, so you were...

Mike Nicholson: Yeah, sorry, on the posher end of B2B now, if that makes any difference.

But the thing that you used to be able to do when there were good relationships between the publisher and the agency, if the agency were putting a plan together and you weren't on it for any reason, you'd go in and have an argument with them about it. You'd go, "Well, no, we should be on it. And this is why we should be on it. And this is TGI data and the NRS data."

You can have that conversation and you can try and claw back some of the budget that you might not have got.

So agentic advertising, where the machines decide what gets bought, that's great from the buy side, potentially. But what if you're a publisher and, for some reason, you just don't seem to be getting any? Who do you go and complain to if you're not getting the budget that you think you need?

Why are we not getting some? Our spend's gone down, let's just say, in one of my old stomping grounds of Telegraph. We're not getting very much money here, and I don't know who to complain to because the machines are doing all the buying and selling. There's no human conversation to have. We should be getting more of this.

Ben Williams: We don't really know what it's going to look like. It's quite transformational what's going on at the moment in general beneath us from a technological standpoint. It's wild at the moment, isn't it?

But Anthropic saying that they're going to make $30 trillion in revenue, which is more than the GDP of China, is just hilarious. I'd be surprised if that actually happened.

But we don't really know what's going to happen with agentic. When I'm working with publishers and broadcasters, this lands really well because I've come up like you, Mike, coming through media sales, so I can talk about it.

Remember the fax machine? You would get a double-page booking through the fax machine from an advertiser you'd never heard of. Do you remember that? Everyone would be like, "Well, that's mine." Ring the bell and you'd book it.

I see agentic as a little bit like that. So you're going to have your direct sales force, and they're going to be with the agencies and the brands and whoever else is buying from them.

The agentic piece is going to be the incremental incoming. So if you've got a seller agent, you're at least putting yourself in the mix to receive agentic buys. Because if you haven't got an agent, you won't get any agentic buys, right? So that's the first bit of logic.

But then there will be human-to-human representations of the agents because something Fern and I have worked on previously, one of the people there was saying that in the US market he was being asked for his agent. So even in the human-to-human chat, people are like, "Great, give us details of your agent."

So it's like human + tech = the deal, so to speak. That's another area which is developing.

Mike Nicholson: So on the buy side then, it's not fully automated by the buy-side agent? That agent doesn't go, "Right, leave it to me", and just goes and does it? So there's an override, a human override? "I don't like that. I want to take that out. I want to put that in."

Ben Williams: Yeah. The position that we're in now is the response from the sellers is the list. Here's the menu based on your parameters, to your point, based on these are all the guardrails I put in and what I know about my brand and what they want and their challenges, et cetera, and the audiences they want to go after.

Whoever's got the seller agents can then respond to that brief. And you've still got an individual on the agency side going, "Yeah, great, this, great. This one, maybe not. We tried that before."

Mike Nicholson: Does it give you options to choose from? Is it like, "That's the plan that we think", or does it say, "You might want to consider this, or this?"

Ben Williams: Yeah. It's the same workflow as traditional media trading. So the brand agent will contain the brief. The seller agent will show what products are available and at what price. And then you can go into a negotiation agent-to-agent and actually confirm the deal.

But there's a lot of human intervention at the moment because we're all testing stuff.

Going back to your point, Mike, about infrastructure, Scope3 have Interchange now, which is really nice. It helps you create storefronts.

What Interchange does is it can help a publisher receive agentic. Anyone that's building on Interchange, for example Scope3, and Scope3 are more focused on the buy side. Anyone that's built a buyer agent on Interchange, if they build a seller agent on Interchange, you open up that marketplace.

The reason why that's interesting is you can either go down one route, which is Prebid and build a custom seller agent, and Prebid have taken ownership of AdCP seller agents and they're doing a great job with it. I'm on the weekly calls with them.

Or you can go off the shelf, like Shopify, take Interchange, plug that into ad servers where it's integrated, like GAM or FreeWheel, et cetera. And it's a much faster way to start receiving agentic buys.

So the infrastructure is coming through. There's not much reporting on budgets flowing through yet, but I wouldn't have expected that. I hear a lot of anecdotal stuff and there's movement from brands. But there's nothing reported yet, and I think that's when people get excited.

Mike Nicholson: Are any publishers missing out today, like right here today, on, "Oh, we didn't get that campaign because we haven't got a seller agent?"

Ben Williams: No, I wouldn't say it's there yet. If I was a salesman, I'd be like, "Yeah, you're missing out. You've got to." But if we're just talking as interested and curious technologists, no. It's more about what's being done at the moment, like revenue tests.

I'm working with publishers. Typically, you do prototyping first, so everyone understands it. They go, "Ah, okay. So a storefront is all of our media kit and all of our data solutions passed as natural language." Then the pennies drop.

Then you go, "Right, next step, let's do a live revenue test." So there is money coming through privately, test budgets like we saw when mobile was forming back in the day. Those sorts of numbers. Carve it through. A little 10K here, 50K there.

I remember doing that in mobile, Jaguar Land Rover with Mindshare, 10K test. We're in that phase at the moment.

It'll either go one of two ways. It will go through the hype cycle and crash, or go through the hype cycle and then the trough of disillusionment.

Mike Nicholson: The trough.

Ben Williams: Gartner, the other trough. And then it all goes vroom. So, yeah, I think as curious technologists, you've got to give these things a crack, otherwise you'll have a bit of regret.

Fern Potter: There is that. It's like being opportunistic in a testing environment whereby the community itself seems really helpful as well. Everybody contributes, it's open-source code, so it feels like a really good environment to be able to test and share and learn.

Ben Williams: Yeah, because it's quite technical.

Fern Potter: What's the expectation from a build perspective? Do SSPs have seller agents?

Ben Williams: They do, yeah.

Fern Potter: As well as publishers?

Ben Williams: Yeah, 100%. In fact, it's the SSPs that have moved faster on this because, as technologists, they've spotted the opportunity.

For example, PubMatic are doing great stuff. They're building on IAB and AdCP standards. And then you've got the new guard coming through. So Aperium Ventures, technologies like Bedrock and those types, who are all experimenting with things like containerisation, which is the IAB's across AMP, so injecting real-time or very fresh data into the bid stream to improve ROAS.

What I love about it is because everyone's almost in a sandbox playing as kids, where everyone's just like, "Woohoo, let's have a go."

I've been in media since '99. I've never known it be so collaborative.

Fern Potter: Yeah, it's great.

Ben Williams: It will close. It will change as people start to make lots of money, and then it will be a closed shop.

Fern Potter: Do you think that's quite a response to the fact that AI is disintermediating revenue budgets, and actually consumer experiences with brands as well? We've just been talking about that earlier today.

Do you think that's also a push into being collaborative in the agentic advertising space because actually it can bring more dollars back to the broadcasters and do it transparently?

Mike Nicholson: I don't know enough about it, but it feels to me, because I always literally got a hat on today, but I always think about this with the publisher hat on because I spent so many years working as a publisher or in adtech as the publisher guy.

I just think that it feels like the control over how much of your inventory you sell is even further out of your hands. I don't know if that's fair.

It already kind of somewhat happened on the open market with DSPs and SSPs. You've got kind of, "Well, let's see how much we got. Oh, we're not quite as much as we needed. Maybe if we tweak the price here", and whatever.

But going back to that point I already made, if it's not going well from a budget point of view, it doesn't feel like you're going to have enough right to reply.

Surely one of the benefits, stated or otherwise, of agentic advertising is it needs fewer people to do it, takes less time, needs fewer people.

So if there are fewer people on the buy side, and there already are since we started selling, like all of us, if that number continues to decrease, it's going to be even harder for the sell side to be able to have a human conversation with the buy side to say, "We should be getting more of your budget."

And if they can't even have the conversation and then the agents kind of direct people in a certain way, I don't know. It's so far ahead of where we are now. It's difficult to imagine what's going to happen, right? But that's the thing that keeps coming into my mind.

Fern Potter: I guess it could go that way. But also I think the direct nature of the relationship between the buy and the sell agents takes away some of the disinformation that's happened before.

So maybe from a less individual perspective, but from a revenue perspective, you should anticipate that you'll get more money for the outcomes that you're generating because there's fewer middlemen, essentially.

Ben Williams: That is one of the theories, where the take rates are transparent and less than programmatic.

Fern Potter: Yeah.

Ben Williams: And also CPMs might be higher as well because it's more drifting to direct. So the theory is higher CPM, greater share of that CPM as it comes down the pipe.

To your point, Mike, about holdcos going through mergers and right-sizing, there's all that transformation happening right now, some for good, some for bad. Share prices plummeting and certain ones dropping out of FTSE 100s and all that.

But then there's a big rise in independent agencies. Maybe that's what's happening underneath us.

I did a project for a client, 35,000 independent agencies in the US. Big addressable market there. You've got big players like Butler/Till and Horizon, who probably people didn't think to go and sell to three, five years ago.

So you've got this new market coming in, and they're very technology-first, but there are people there.

You know, Butler/Till, the CMO, she's always talking at events and she's very active. And then there's a whole team behind her executing.

Whereas the holdcos, yeah, like going out and taking the trading team at GroupM for lunch and all that jazz, and your buyers of VW Group and Sky, those days might have changed.

But then there's a new guard coming through with some great budgets because I think there's a great pitch for independent agencies over holdcos. I'm not saying one's right or one's wrong. There's a lot of bad press recently about rebates and stuff.

That's probably going to turn big brands off. Like, hold on a minute, if we're going to hand you two billion euros of media money, we want to make sure it's being managed well.

So maybe independents are getting a good bite of that cherry.

I was speaking to Dan, as I said, at Heineken earlier. He used to work at MediaCom about three times, I think. And now he's in-house at Heineken.

It was refreshing to hear Heineken's in-house approach, and it kind of differs from the holdco approach. They're kind of testing and learning stuff all the time. It's the way it used to be. You used to be able to go into a big agency, "Oh, we've got this new bit of tech", and then they'd try it for a bit.

But now it seems to be that it's not a closed shop, it's much harder to get a test or any sort of budgets or anything going. They're like, "No, we use this DSP for our video and this for our display." That's all kind of set up.

Mike Nicholson: And to your point on independent agencies, I think things are moving too quickly for that to be ideal, right?

Because you look at all of the SSPs, let's just state SSPs. You look at all of the SSPs and you made a decision today for the next 12 months. In three months' time, it might be the wrong decision. It might be the right decision today, but in three months' time, that one's leapt ahead and that's much better for you now.

If independent agencies are willing to go, "Do you know what? We'll just switch over to there", because we're agile enough and small enough to be able to do that, whereas the holdco goes, "Well, we've got a trading deal. We've got to carry on going", then there's potential risks there, right?

Fern Potter: Yeah, there is. Twenty years of my career were spent at the holdcos and I adored it.

When I look at it now, I've got a sense of sadness because I think it has definitely changed. And to your point, there are a lot of amazing people that still work within those businesses, but it's the commercial rigidity that exists based on those deals.

That's what we've seen in press that I think kind of harnesses the entrepreneurialism, the test and learn. Those budgets don't exist.

I know this from being at Multilocal for the previous three years. Trying to have a conversation within the agencies was really tricky. But the independents, and I don't know whether you can even call them independents anymore, I think there's this whole new...

Ben Williams: It's a new guard coming.

Fern Potter: Should we give them a name?

Ben Williams: Yeah.

I was speaking to somebody that worked at one of the big holdcos that you know. I won't name them on here, as they were leaving the holdco.

I said, "Oh, I'm working with this client, an adtech client. Who should I speak to about trying to get..." They were like, "I wouldn't even bother. You'll just spin your wheels for six to eight months and nothing will happen."

I was like, "Oh, cheers. That's really uplifting."

They were being honest because they were leaving, right? And they were like, "It just won't happen. I can give you some names if you want, but you'll just waste loads of your time."

Fern Potter: I think decision by committee as well. You've got so much siloed within the holdco, so maybe that's another key difference with independents. There's a lot more cohesion.

I've worked with the Butler/Till team as well, at Multilocal, and Amanda and Scott. They're fantastic and they're entrepreneurial and they've got depth of knowledge of their clients, but also their own organisation across all functions.

It's not like there's a team here, a bit of business there, that there. I think that cohesion and the strategic direction actually lends itself to being able to move a lot quicker and then test in the right way.

So, yeah, maybe the displacement actually is going to come on both sides, from the buy and the sell side.

I'd definitely love to see more of the broadcasters and publishers investing into being able to set up the seller agents as opposed to it just being the SSPs.

I know that's tricky because there's a lot that goes into capability, capacity, hedging the bets. I've got a list of problems. Where is it on my list? Where do I invest my time as well as my capex?

Mike Nicholson: But having worked at a publisher for many years, there's a dev queue. You go, "How much money is this going to make us?" And it gets put in the dev queue based on how much money it's going to make you.

At the moment, if the answer is "not a lot", it might do in the future, but the answer now is not a lot, there's probably something that they need to work on now that will get them more money in the short term.

Fern Potter: The immediacy. It's funny, I've come across that at IA as well, where we've kind of gone, "Oh, we could set up these and we can work in increments, work with you to get quick wins in to demonstrate the prioritisation."

But we get a closed door: "We've got a 12-month locked-in dev queue. Come back when we've got rid of this tech debt that we've got to change because we've got to do that."

I completely respect the decision, but I think in 12 months' time, you're going to be in a position where you're left behind if you've not tested now.

I'd say that's probably, I mean, we talk about some of AI broadly as well, but I'd say that from an agentic advertising perspective.

Ben Williams: There's a new universe opening up, though. I think one of the things I have to do is keep a few different bets. I can't just be all in on one because it might not happen, it might not scale.

But you have to do these small, calculated bets.

I didn't go all in on, well, I didn't even go in on NFTs and crypto and all of that stuff when we all went temporarily mad during COVID, thinking that everyone's going to be transacting with crypto in five years. We're not, and we didn't. "Oh, we're all going to be in the metaverse." And we're not. We're not going to walk around with headsets on.

But what is happening, what I am seeing is, one business I'm working with at the moment is a sports data business. I won't go into detail, but there's a whole new universe of supply opening up through things like creators.

We've come up through traditional digital publishers, which is your print-to-digital title and how that got monetised. But there's a whole new estate opening up where the eyeballs are going.

If you talk to those companies that aren't encumbered by programmatic and legacy ways of doing business, the attitude is very different. It's like, "We want to. We're thinking about the future and we're thinking about young audiences, where the eyeballs are."

One in particular has got some big global brands attached to it, and so it's a great opportunity.

You see the perspective. You see legacy digital publishers who are really struggling because Google traffic's been hit, their programmatic stack hasn't been looked at in 10 years and they're struggling.

And then you've got this new guard coming up in different categories, and they're more likely to experiment, probably push the envelope for the next chapter.

Because I think we've gone analogue to digital to AI. I think that's it. It's like the skip from analogue to digital. We're kind of seeing that digital-to-AI hop at the moment.

If you remember what it was like doing analogue to digital back in the late 90s, early noughties, we were selling skyscrapers on websites at £60 CPM, and it was just static. We were like, "Really, you're going to pay us 15 grand for that?" What we just stitched into a website.

We didn't know what it would become. No third-party tracking. Just send them an Excel sheet. "There you go, there's your delivery report."

Mike Nicholson: No, yeah, it was actually that. There's a delivery report.

Ben Williams: Yeah, it was actually that.

Fern Potter: Funny enough. God, that takes us back, doesn't it?

Ben Williams: People look at me a bit strange when I speak to young people about that. They go, "No third-party trackers?"

Mike Nicholson: Trust our Excel spreadsheet.

Ben Williams: Yeah, it's absolutely fine.

Fern Potter: I love this nostalgia. I found a deck from 2017. That's not even 10 years ago. And it was about transitioning budget from analogue to digital. What percentage of budget should you put into digital advertising?

I was like, how are we even having that conversation 10 years ago? That feels like it should be a lot longer.

But it was like, yeah, you should invest. And it didn't even say digital channels, it was just digital channels. It didn't tell you the brand or anything.

But, yeah, it was quite wild to me, the different levels of expectations and maturity.

Then it goes back to your point, right? It's novel. We're in our infancy. You can test, you can learn, but nothing's quite certain. So I think there's a level of people needing trust and confidence in not only the process, but the people within that process.

Ben Williams: Governance and trust are so important.

IAB is an interesting one. Obviously, it's a member organisation and some members pay more than others. You'll have some very big names in there, so there's agendas and politics.

And then you've got others, like Prebid or AAO, who seem to be a lot more, I don't know, transparent and thinking about the future of the internet and the future of the industry.

You've got some big agendas out there. Meta have just coughed up 18 billion to settle what was quite a dark lawsuit. So there are some bad actors out there, right?

But humanity is always going to have an interest, I would imagine, and the internet will always grow. And the money that can be made from the internet will.

Advertising has grown 20% year on year for 20 years. It's one of the only industries that has grown that amount. And I think we'll see another spike in the next few years when it's like, "Oh, there's a whole new way in which we can monetise it in a really wholesome way."

I think browser behaviour will change as well. Devices will change over the next few years. You'll see all this transformation happen because I think what we're doing at the moment is we're kind of trying to have this AI experience on previous chapters of technology.

You'll see new browsers, new hardware, you'll see all that coming through. And then there'll be, I think, really good monetisation routes for that.

Fern Potter: I love it.

Ben Williams: Awesome.

Fern Potter: Sorry, no, go on.

Ben Williams: No, I was going to say...

Fern Potter: I just love that. I was going to ask you about the brands or even the businesses that you're speaking to because I think we're not AI-native. The products aren't, the hardware isn't, the software isn't, the businesses aren't.

This transition into agentic AI that we're going through, there's friction, it's wonky. There are elements of should it happen, could it happen, what does that mean for me? It's very personal.

But do you think actually now, we're iPhone-native or we're search-native, as it was? Those are all things that we just kind of got used to over that period of time.

I just wonder, I think we'll get to that point. Are you seeing businesses that are already kind of AI-native, that they're new, that are coming into agentic advertising?

Ben Williams: Yeah. There's one that I can't name.

Fern Potter: Intelligent Assistants.

Ben Williams: Well, you are. You guys are. I was including you, including you and Sarah and James.

There's a technology business literally AI-first, and now thinking about what monetisation could look like. And that's fascinating, actually. It's like helping figure it out.

Then you've got some really good, I mean, the people at Bedrock, which is AI-first, they're doing brilliantly. They've got a great team and closed their seed funding this year and are just open to experimentation.

They're trying all the protocols. I think people want to work with them because they're like new kids on the block. They're not a client of mine. I'm hoping to partner with them on something.

So it's opening up a little bit.

But then you also work with clients who are legacy digital businesses and you go in and they want to do all the AI stuff.

What usually happens, well, we've got a process. You go and you look at the data and you go, "This is the data work that you need to do to begin."

And people go, "We've got to do what?"

And you're like, "Yeah. Do you not understand that the fuel for artificial intelligence and machine learning is data? And the data needs to be stored in a certain way, which means one might have to migrate databases."

And people go, "Whoa, hold on a minute. I thought you were just going to build me a chatbot."

Fern Potter: Yeah, but this is it. It's just like, cool, go and build a chatbot. Nothing wrong with that at all.

Mike Nicholson: It's businesses that are using it as a tactical add-on rather than a strategy change, right?

Ben Williams: Yeah, it's the different mindset.

I think this is why our partnership's great because we're complementary. We'll do the boring engineering stuff.

Fern Potter: I don't agree with you saying it's boring. I think it's transformational.

Ben Williams: The transformational work.

What I like about, and I know you're still early in Intelligent Assistants, but in fact, question back to you, Fern. We've worked together on a few things so far, and you know how we work. When you're talking to clients, how do you pitch IA?

I know you tend to focus on the SMB market, which is a great market segment to focus on, by the way. What is your positioning when you're talking to clients? I don't mean to put you on the spot there. Sorry.

Fern Potter: I talk about myself and IA. Absolutely.

For us, it's about using AI to inform human intelligence. So we want to work with businesses whose objective is to grow, either from a revenue perspective or increase their capability and productivity within their current teams.

We go into a business and we've got a straightforward process, which is think, build, adopt.

Think is finding the frictions that exist within that business that are fundamentally barriers to growth. In doing so, we look at what the success measures of AI application would be.

We build intelligent engines. So it's not just about using AI. There are also APIs within that, there are automations, understanding the team workflow, understanding what knowledge sits where and unstructured data.

On most occasions, I'd say that we're looking at the same amount of qualitative as well as quantitative data that goes into feeding the engines to give them the right context.

Then we build that engine to fix that business problem so that that business can grow.

The final part of it is adopt, which is the most important part, which is the change management.

So it's not just using the tool, but it's also working with the individuals within that team to go, "This is how it significantly changes your role and shifts that direction."

Putting those guards in and then that constant measuring of, is this still working? Is it hitting the goals that the product was meant to do?

So very much custom solutions, but have a sweet spot where there are businesses that have, and I mentioned this in a conversation before, I think entrepreneurial C-suite, entrepreneurial CFOs, I think, is going to be my new target market, where they're looking at the revenue health of the business and saying, "This is how I could grow, and I'm not."

As opposed to just looking at a P&L with an FTE and a cost line and going, "Well, I don't want to invest in this now."

So, yeah, it's a real mixed bag at the moment in terms of the clients that we're talking to. I think the opportunity to be able to bring in the intelligence layer with the AI to develop the business...

Ben Williams: There's a great story from about three weeks ago. Me and Fern were meeting in Manchester because we were on a podcast with the Dean.

We had a pre-meet, and Giovanni, who works with me and is based in Leeds, we had a nice little couple of hours.

Obviously we won't name a client. You were like, "Oh, I've got a brief that we could do a bit of it, but it's more for you." And you said who it was.

I was like, "Oh no, I've been briefed on it as well."

It was like, "Well, hold on a minute. Why don't we do a combined review on it?"

Because what I can see is the stuff that you do, we don't tend to touch that stuff. Not because we don't want to, it's just that we're kind of more focused on the deep transformation work.

I'm not talking about seller agents and things like that, but we're more interested in the engineering piece.

Then I could see straight away, it's just like, "Oh, hold on a minute." And then you're building on top of that.

We put a pitch in for that. I think that's also really good. English or UK-based tech businesses collaborating to have an independent voice in the market.

Fern Potter: Sovereignty.

Ben Williams: Sovereignty, yeah.

Fern Potter: UKAI is the sector trading body. They really are promoting and supporting sovereignty, which you're part of, aren't you?

Ben Williams: Yeah, we're part of it.

Fern Potter: I think that's fantastic because there's so much innovation that's happening in our country. There's a lot of opportunity for us to build and grow and learn from each other.

Those kinds of collaborations, I mean, we have a partnership as well, right? Bringing everybody together to help move things forward.

I think there's something in ensuring people don't feel left behind in that progress as well, if that's what collaboration can do.

Ben Williams: I agree. I think we'll just unlock so much opportunity and hopefully let it all... We all just want to grow the economy, I think, as well. At the end of the day, create jobs, not take them.

Fern Potter: Yeah. Together we are stronger.

Ben Williams: Yeah. We're all Brits, right? So we all want our country to do well.

I think that collaboration piece, not just with UK companies, but what we can do together on the international stage as well, like what we're talking about in APAC.

I'm sure this meeting will probably get some ideas cooking and we'll think about how we might collaborate.

We are leaders of our industries and what we do actually counts.

I was talking to a good mate who's senior in government. He's part of the Civil Service and he's a good mate of mine. He's responsible for budgets and whatnot.

I talked to him about what I'm doing, and he was like, "Don't worry, I've already checked you out."

I was like, "What do you mean, brother?"

I got to know him in Brighton. He's a really good bloke. I was joking with him, I was going, "For all you know, I might be a spy."

He genuinely said this. He went, "I've already checked you out."

I was like, "What do you mean by that?"

Fern Potter: Was that before or during the friendship?

Ben Williams: I think he was basically just doing a little vet of me because I was talking about what I was doing.

But then he takes an interest in what I do. We underestimate how important all of these collaborations and pitching and growing the business and paying our corporation taxes and the case studies are. It's really, really important.

Through the collaboration, Fern, you and I have been collaborating a lot this year. We haven't announced our partnership because we don't need to because we're just doing it, aren't we? Just cracking on.

But there's a lot of these great collaborations that are starting to happen now.

I think we shouldn't be an island, the independents, because you're either working for a startup, scale-up, or you're independent, or you're working for one of the big platforms, right? You're working for Google or Meta. Great. You're on your big six figures and your RSUs and all that stuff, but it's US IP. You're developing US IP. There's nothing wrong with that.

But what we're doing is we are collaborating as independents to try and do something which is a force for good. I think it will be positive in the long run.

Fern Potter: Positively, I see there's something in that as well. I think it was back to a point you made earlier, Mike, that if you've got collaboration and partnerships, does that give you more access to revenue that you might not have seen as well?

Because we were talking a bit about there is quite a lot within the vendor space now. And that's not actually just individual businesses, that's tooling as well, isn't it?

It's like almost services are software, as opposed to software as a service.

Ben Williams: Services are software.

Fern Potter: Services are software. Thank you. It's not just me.

So there's loads of tools, there's loads of individuals, there's loads of independents.

If you can work with businesses that pre-collaborate behind the scenes, go, "Right, we've got this brief. Here's a collaboration", got permission to share that with individuals and come back with a collective and go, "Here's how we can answer that with the tooling, with the right people, individuals and companies behind it", does that actually then give more people access to that slice of the pie?

Ben Williams: Should do. The more people you collaborate with, the bigger your second-degree network is, to use a LinkedIn terminology.

But highly relevant, complementary skills enable you to help drive businesses forwards, I think, which is exciting.

Fern Potter: So to that point, who's driving agentic advertising forwards? You don't have to name names, but are there any specific verticals, any specific trends, or is it too early?

Ben Williams: I would say the brands that moaned the most about the murkiness of programmatic are the ones that are going first.

Fern Potter: Interesting.

Mike Nicholson: I was going to say to you earlier, you said about bad actors. There's always been bad actors in programmatic advertising. Is agentic buying going to make that better or worse?

Ben Williams: Well, where there's money, there'll be bad actors.

Mike Nicholson: Rogue seller agents.

Ben Williams: Yeah. We'll have to deal with all this. I don't know.

In my head, I've got something like Space Invaders as your console if it's getting the rogue...

There is some great technology, ads.txt, sellers.json and brand.json, and the governance around it to protect against fraud and things like that.

But there are all sorts of working groups looking at this.

If you sit on, I won't name names, but if you sit on the Prebid weekly strategy call for agentic, it's literally engineering from top adtech media businesses from around the world.

No one's coming in with the agentic... It's literally, do we protect the health of what it is that we're trying to do? It's a real custodian-type attitude, actually, which is fascinating.

But if it scales, there'll be bad actors, and the bad actors will be the people that will be sneaking take rates, high take rates, and finding ways to kind of hack the system to further their agenda.

But to answer your point, nothing's been properly reported yet, but I know anecdotally the names of some of the brands, and it's typically the brands that were annoyed about their revenue disappearing.

A lot of these big global brands have KPIs written into their teams, which is, you know, 10% of your budget has to be agentic.

Because the boards are like, "We need to be experimenting with this." So we're going to motivate you to experiment with it.

Fern Potter: I love that. It goes back to my, you know, 10% of your investment has to go into digital or something like that. But it's that evolution, as you said, from digital into AI.

I think we talked about the word programmatic a little bit as well there. You can often get in your head that that's just maybe display and video.

But actually, you've seen all of TV going programmatic, digital out of home and audio as well.

So I think we have to also imagine that this is a world where agentic advertising covers everything you buy, not just programmatic, as that could be too much of a siloed view.

I get the bad practices in programmatic might be a lever to have wanted brands to want to invest and test in agentic advertising.

But as that kind of programmatic mechanic takes up more formats and has greater breadth, agentic advertising is going to take over all of that media buy, essentially.

Ben Williams: Yeah, I would say the opportunity to... I shouldn't say it will. I just see it as a way to make existing teams more powerful.

There will be jobs. There'll be new jobs from all of this. It's not doom-mongering. People will have to be operating and building and negotiating.

So I think the new generation is coming through.

Giovanni, who works with me, his first entry point into media is building seller agents for clients. You know what I mean? It's hilarious.

My first entry job was making 60 calls a day, 18 of which had to be effective, to small hotels up and down the country to see if they had a conference suite that they wanted to advertise in Venue, which was a Haymarket directory.

And Giovanni, his first entry point to media is literally setting up seller agents.

Fern Potter: So, you know, he's really winning there, Ben.

Mike Nicholson: 60 calls a day.

Ben Williams: Yeah, that was the big deal.

Mike Nicholson: And your sales manager at the end of the week would check your numbers versus the call logs.

Ben Williams: Your call logs, yeah.

Mike Nicholson: I think that we've kept you far longer than we said we would, so we'll wrap up soon.

Last question for me, I don't know if you've got any closing thoughts as well, Fern, but how is it going to change the shape of the Lumascape?

Are we going to end up with buyer agents talking to publishers' seller agents and everything else in the middle kind of goes away eventually? Or is it still going to need a DSP and SSP, a data platform?

Ben Williams: The agents that are being built right now are brand agents, measurement agents, signals agents, creative agents and seller agents. Reporting is within that as well.

So that's the constituent of the transaction, intent through to activation.

I think everything's being built on the existing infrastructure that we've just spent the last 20 years building, right?

Think of your fibre optic going into your home. It's using the existing infrastructure, and then the infrastructure is bit by bit upgraded by Openreach and BT and whoever else.

Same with the progression of media.

What will probably start to happen is people will go, "Are DMPs full of junk? Do we need DMPs? What does a CDP do?"

And then what will happen is, as the money starts to figure all that out, usually, I guess, the biggest budgets first, the ones with the multi-billion euro, dollar, pound budgets will start to go, "Well, we're actually starting to send our money, and you need to have this."

And if you're going to do an RFP to us, and if you're in an MSA negotiation, you're going to have to.

The change usually comes in, and I'm sure you've seen that working for holdcos, the demands coming in from the big global advertisers.

I'm sure you saw it in your agency days, like, "Oh, this is new coming into the requests and the asks."

Fern Potter: Yeah. "Oh, we need to have a measurement partner."

It's very much that conversation of, if you can, should you? And if you should, can you?

As you went on, the contract negotiations got longer and longer.

I think I spent three months negotiating the Jaguar Land Rover contract at Discovery because of media quality.

Because Moat launched and you could start using viewability. So how did that impact every country, every ad you bought globally? What were the regulations? It was fascinating.

But yeah, to your point, there become more and more principles that you add in and more guardrails, I guess, for the brands.

I think that Lumascape will contract and expand in other ways.

There's always, as you said, when there's money, there's always an opportunity to build something to capture some of it.

We're very good at optimising problems, not solving them, because that's how you make money. I'm not saying we personally, but as an industry.

I think that's why that Lumascape's always so huge.

Ben Williams: Yeah. Terry Kawaja always says that. I don't know Terry, but when you see him on stage talking about the Lumascape, he's like, "It's ripe for consolidation", which obviously he's got a vested interest in that because he does M&A, right?

But yeah, I think there are two answers to your question there, Mike.

One is what the brands, when they come around to their agency reviews, say: "Oh, we want you to have this." Because the boards have gone, "Do you know what?" And that dictates the winners and the losers over time.

And then the consolidation piece. I think there's going to be loads of M&A going on over the next few years.

I think there's quite a lot of dry powder because this year was particularly weird. There was a lot of capital expenditure that just froze because of war and tariffs and all that stuff. Literally just froze.

But I think things will start to, people are going to be like, "Shit, we need to get on with it."

So there'll be new entrants, there'll be consolidation, there'll be losers. And I think acronyms will change.

I was talking about this with my engineering team because we've got some white-label technology that we're selling to clients sometimes that they've built over time.

They call it ATS, ATD, agency trading desk, DMP, SSP.

We were talking about it. It was like, what is it? And Andrew was like, actually, a lot of these acronyms are just going to change over the next few years.

Mike Nicholson: Don't mean the same thing.

Fern Potter: Yeah. Or the functionality is different as well.

And I guess through that M&A process, you've got so many different operational ways of working.

If you look at Salesforce, it's like, it's CRM. It's not anymore, is it?

Ben Williams: No.

Fern Potter: Agentic workflow. What do we call it? Services are software. Got it right the first time. But it's a whole suite now, and it's bringing all those building blocks together.

Ben Williams: It's a data business, isn't it?

Fern Potter: It's a data business.

Ben Williams: They've got Slack now, so that's embedded within organisations. They're an AI-first business, though. Even though Salesforce is, what, 20-plus years old, now they are completely pivoted towards being AI-native. So they'll be a winner, I would guess.

Fern Potter: It's funny you talked about Terry because he did put an AI Lumascape out quite a while ago. I don't know whether I've seen it updated.

Mike Nicholson: I forgot who was on it.

Fern Potter: Yeah, me too. It was really early doors. And I remember him just saying, obviously, this is going to change a lot.

But I wonder if that, I mean, maybe it just moves too fast. Or maybe there's a product we could start building something.

Ben Williams: It's interesting to see who FirstPartyCapital and Aperium and a few other funds are invested in because they're doing all the due diligence.

When you see them go, "Right, we're going to back a couple of million dollars into this business and lead the round", you're like, they're the ones to watch. They're the ones that are coming through.

Fern Potter: That's a good indication.

Ben Williams: I think Aperium have got a whole bunch of recent successes. Bedrock's one of them, and Scope3 is in there as well.

I think Brian O'Kelley and his co-founders are doing really solid work on advancing the industry.

I think it's going to be an opportunity for people to make a lot of money. I think there will be losers, but that's just, as we all know, media is a tough industry to be in. You have to stay relevant, you have to stay networked.

I'm glad we're not recording this in Cannes. Can you imagine?

Fern Potter: With a straight face and everything, it's really hard. Rosé, anyone?

Ben Williams: Yeah, well, I think there's a lot of people that do very well.

Mike Nicholson: Yeah, I mean, let's face it. Staying at the Majestic in the suites. We've all worked on that same experience.

So what's your big bet then? If we're going to end on that, what's your big bet in agentic advertising? You can only have one.

Ben Williams: I would say that browsing behaviour is going to change.

What you will see is everything start to make sense as the user experience changes.

I think what's going on now is we're part of a massive society, humanity, and there's a lot of us looking at this. We're kind of crowdsourcing all in real time, from your Anthropics through to your agentic advertising startups and everyone in between.

The infrastructure is changing, the hardware is changing, the software is changing.

I think in about two to three years' time, I think by the end of 2030, smartphones and browsing will have changed, and then agentic will make a lot of sense for that new paradigm.

So I think everything's sort of happening.

If you remember, if mobile was Steve Jobs with the iPhone and the App Store, that was one individual. Social was essentially Mark Zuckerberg with Facebook. And then social and mobile was the big part of the last transition.

This one seems a bit different.

Of course, you've got your heads of the LLMs, your AI businesses, Dario at Anthropic and Sam at OpenAI. But this one feels a bit more, I don't know, dev-community-driven.

So I think monetisation of that. I think the internet is going to be much bigger than it is today. I think the opportunity and the addressable market is going to be much bigger.

And I think one thing it will unlock, it'll unlock two big things.

One is brands will have more confidence in investing in it. I think that's one big unlock.

And the second big unlock, which will be, I think this is why you've launched IA, is the unlock for SMBs.

SMBs are 60% of the UK economy, right? So let's assume that it's 60% of the global economy. Whoa.

If you're going to unlock that into advertising on the internet, doing the internet properly and easily, agentic, I think, will be one of the unlocks for that.

So that's my two sort of big bets. How we get there will be a bit scrappy. Some things will work out. Never straightforward, is it?

Mike, you got any bets after our conversation on agentic advertising?

Mike Nicholson: Honestly, this is the end of my fifth podcast. I don't think I've got any thoughts left.

But it's been a really interesting conversation.

Have I got any bets? No, I'm not deep enough into it. I'd just be guessing.

Keeping an eye on what happens over the next kind of 12 months.

I don't know if we're going to reinvent the problems that we had with programmatic around bad actors, rogue seller agents, publishers getting squeezed more.

Publishers always end up paying for this somewhere along the line, which is quite a negative way of thinking about it.

Fern Potter: But you can be healthily cynical.

Mike Nicholson: Let's go with that.

Fern Potter: Or is it a healthy dose of scepticism?

I think that helps because that actually then helps people talk more, justify more, measure more, share more. That's why that collaboration is so important.

Mike Nicholson: And Fern, what are your bets?

Fern Potter: I didn't prepare one in advance.

Ben Williams: Wing it.

Fern Potter: You know, I'm on board with it. I am so, you know, not just AI. If I keep my head and narrow my focus to agentic advertising, I absolutely adore it.

I think it's transformational. I think it unlocks a lot of opportunity for publishers, broadcasters, I'm trying to think of media owners, shall we say, that whole collective that wouldn't have had access before.

I think it's going to be a door opener into new revenue.

And I think it's also going to be an opportunity for the buy side to open up new avenues of advertising.

Because I think we talk a lot about, as I said, programmatic, which is digital, which kind of, you always think in your head, you go to broadcasters and publishers.

But I'm thinking out of home, I'm thinking agentic advertising on, I don't know, Airbnb TVs, I'm thinking digital screens, sports venues.

Because people are looking at utility ownership, because AI is changing the way that we consume sports and media and stuff.

I think that has the agentic foundations because it's going to be agentic-native.

We might not get there now, but that's what excites me as to where we're going.

But I want to be really respectful, I think, of what you said in terms of the bad actors piece, because I think I've found that the most, I don't want to use the word upsetting because it's quite emotional, but the most disappointing part of being, and I've been in programmatic since the inception of programmatic, is exactly that.

Hopefully, let's learn.

As Ben said earlier, where there's money, there'll be bad actors, unfortunately, in any industry. Bank robbers, highwaymen back in the day.

But there are bad actors, and then there are those that manipulate the situation to make money.

And I don't think that there's bad, you know, the bad, bad actors, for want of a better word, or those that are bad...

Mike Nicholson: How bad is bad?

Fern Potter: Yeah. I know it's a bit of a spectrum. But it's a bit naughty.

"I'll pay for this because it will help you", and it doesn't help.

Those kinds of operations that popped up because of programmatic and took the take rates and took the investment away from the publishers, which we know has had a negative effect in terms of how they run their businesses.

I think that's such a cautionary tale that we need to take forwards.

Mike Nicholson: Yeah. There's always been, maybe understandably, but there's always been an over-focus on the buy side of the industry and less focus on the sell side.

The problem is in this new AI world, as I can see it, is that the content that publishers produce is feeding this whole thing and they're not getting very much money out of it. They're not getting potentially the fair share.

To your point with the PwC, it was 50% then. Goodness knows what it is now.

Somebody, a tech startup I spoke to this week in the US, is trying to solve that problem. He reckons it's about 38% now that publishers are getting, so 38p in a pound in some instances.

And also it's the CPM as well, the programmatic CPM.

If you look at the reports on the publisher side and you see the CPMs, and then if you go and chat to someone like, "I booked that for like 20", it's like, wow, that came through as two.

Fern Potter: I know we wanted to wrap up, but I just want to mention on that. I think the paradigm shift into who has the power could also be disrupted by agentic advertising based on the fact that, from a retail media perspective and brands owning data, they are becoming media owners as well in their own right.

So one would hope that when, for example, I'm IKEA and I'm selling my data to other brands, I'm IKEA and I'm selling a media experience in the displays within my stores on the televisions, will they then see that kind of sense of the lack of parity on the buy side and sell side?

And actually maybe that's where agentic comes in.

Yes, I know it's buy and sell now, but perhaps that does collapse the model, and there becomes this more equitable, circular way of working because the brands are the publishers, the publishers are the brands, and there's a bit more of this collaboration.

Ben Williams: Yeah.

Fern Potter: And then maybe we unlearn some of those bad behaviours.

But I get where we are now. I've just got this feeling and hope, I guess, that that fundamentally shifts the foundations of agentic advertising.

Ben Williams: On that point, there are two organisations that I'm working with at the moment where they're both buyers and sellers of media. So maybe it's already starting to crack open a little bit.

Fern Potter: Yeah. And that being in the power of the brand or the media owner in itself, as opposed to being a platform.

Ben Williams: Yeah. Because, you know, no one wanted the buy and sell side under Google, for example, because you're like, "Oh, you're selling it and you're buying it", hence the DOJ case.

But if you can have that within a branded environment, that's actually better for the customer.

Fern Potter: Yeah.

Ben Williams: Consumers want to go to, because when they've got 10 minutes and they're chilling out and they're on their device, they want to be entertained. Or they're buying something, or they're researching a holiday.

Mike Nicholson: My other big bet, which I've just developed while talking, is that brands will spend more money on trying to be the content rather than the advertising.

If you think about how LLMs go and answer questions and come back, they're already furiously trying to figure out how they become a part of that answer.

I've talked about it on a previous podcast. I've done five now, so I'm trying to think, did I speak about it on this? Was that this podcast?

But there was a crazy stat, a medical stat, where people were asked about medical data, and 81% of people said that they trust ChatGPT medical data as much as a GP.

Ben Williams: ChatGP.

Mike Nicholson: Crazy.

Ben Williams: Catch that. GP. Crazy.

Mike Nicholson: Very quick of him, ChatGP. Trademark it.

But I think you're right.

Fern Potter: And I can, even actually through the lens of curation, I could see that happening.

Brands going and segmenting their data and pushing that into the SSP to be curated on. Or creating content that the LLM will find and then surface.

Mike Nicholson: It's kind of like a clean room, right?

If you see a brand say that something's really good about themselves, you think, "Yeah, you would say that."

If ChatGPT says it, it feels like all of the brands have been stripped away. Now you're just information. And people really trust it.

I've had new

‍