Welcome to the Agentocene
Andrew Mole
16 July 2026
|
Agentocene (n.) The current era, defined by the emergence of autonomous AI agents capable of perceiving, deciding and acting across complex systems without human intervention at every step. Wider than any single industry — spanning logistics, finance, healthcare, software engineering and beyond — the Agentocene represents a structural shift in how decisions get made and work gets done. For advertising: The Agentocene reshapes advertising at an architectural level — not by optimising the existing stack, but by removing the intermediary layers that only existed because humans needed interfaces between systems. |
Every technological era gets named eventually, normally a long time after it has happened (for obvious reasons). Whilst we are officially in the Holocene, you are as likely to hear the term ‘Anthropocene’ thrown around as an acknowledgement of the outsized human impact on our world.
To this, we would like to acknowledge the current shift in emphasis in lived experience and impact on society and the world beyond: this is not the AI age. Not the data age. It’s the Agentocene. And if you’re still framing your business strategy around anything less, you’re likely already behind.
Naturally, we view these changes through the lens of the Advertising industry, though the impact is holistic.
What an agent actually is — and why advertising needs one
Agents aren’t smarter software. They’re a different category entirely — systems that perceive, decide, negotiate and act autonomously across complex, dynamic environments without (requiring) a human sign-off at every step. That distinction is significant, because advertising infrastructure was practically designed to be disrupted by this. Every layer in the stack exists because humans require(d) interfaces between systems. Agents don’t need those interfaces.
The stack was never the answer — it was the workaround
Programmatic Advertising was a theoretically brilliant solution to a real problem. But it calcified into something that serves the infrastructure more than the participants. DSPs, SSPs, and the multitude of intermediaries that sprang up in between were built not because they created value intrinsically, but because they bridged a gap that humans couldn’t otherwise cross at scale. The platforms connected the pipes, and the layers between kept them honest, but all in pursuit of increasing flow efficiency. Over time, that pursuit became less about innovation and more about self-justification. And now, those gaps are closing. The bridges don’t disappear gracefully. You have to choose to remove them.
The economics work for everyone — when you remove the middle
This isn’t just a media owner or an advertiser story. A meaningful chunk of every media dollar evaporates before it does any productive work (the work undertaken being to ensure good actors throughout the chain) and paying tolls to intermediaries that sit between the brief and the impression. Advertisers get less reach, worse signal and diluted ROAS. Media owners get a fraction of the value their audiences and environments represent. Stack consolidation through agents fixes both sides simultaneously. Better returns for buyers. Better yield for sellers as a direct result of removing legacy structural waste.
That both-sides win is real, but it’s worth being clear where it starts. The supply side has been the underserved end of this market for years, and it’s the side we chose to build on first. That heritage is exactly why we can now build credibly for both sides of the transaction — an agentic stack that genuinely rebalances the economics has to be built from the supply side up, not retrofitted by the players who profited from the old one.
Data is the engine, but only if you can actually use it
Every serious advertiser and publisher has data. The gap between having it and activating it at scale, in real time, across fragmented environments, is where most of the value gets lost today. In an agentic system, advertiser first-party data, publisher audience signals and contextual intelligence all become live, combinable and usable at a scale no human-operated stack can match. The data was always there. The infrastructure to activate it properly wasn’t, due to the limitations of the humans that were using it.
What the agentic stack actually looks like
Buying agents that evaluate inventory, negotiate deals and activate campaigns directly — informed by live data signals. Selling agents that manage yield, packages and demand relationships autonomously. It’s not a layer on existing infrastructure. It is the infrastructure. We are finally able to consolidate the outcome and ‘platform’ into a closed loop. This is no longer arms length, this is software in service of a result.
Beyond programmatic: the deals that were never automated
The programmatic stack, for all its complexity, only ever addressed one part of the market.
The high-touch end: homepage takeovers, content sponsorships, influencer partnerships, premium placements are still done the old way. Phone calls. Emails. Excel spreadsheets. Relationships managed manually at every step, with all the friction, delay and margin loss that entails.
Agents change this too. The same autonomous systems that handle real-time bidding at scale can handle the negotiation, execution and reporting of deals that have never touched a platform. For the first time, the full spectrum of advertising transactions, from programmatic to bespoke brand partnerships, can operate with the same efficiency, data activation and speed. The automation of what was always manual isn’t a nice-to-have. It’s where significant value has been stranded for years.
pubX’s thesis: the stack gets smaller, and smarter
We’re not building tools to make existing platforms more efficient. We’re building the Agentic layer that makes them unnecessary, following the logic of what agents actually are, applied honestly to how advertising works. If an agent can do what a DSP does, better, faster, cheaper, with full data activation, the DSP is a cost centre waiting to be removed.
But this is table stakes: a better economic flow is what will enable Agentic Advertising’s foot in the door, but we can also use Agentic to develop a better advertising system, using the almost infinite capacity of Agents vs. humans to optimise to outcomes in previously unimaginable ways.
The Open Web needs this to work
There’s a bigger argument here. Quality publishing, the kind that creates real audiences, real context, and real brand value, needs a sustainable economic model. So does the content layer that trains and informs AI. Agentic stack consolidation isn’t just good for margins. It’s how the open web gets a fighting chance.
Beyond the Open Web, the same blueprint of economic improvements and better results can be transcribed to all forms of advertising.
The question for everyone in this industry
The Agentocene is here. The question isn’t whether agents reshape advertising infrastructure, it’s whether you’re building for the version of the industry that’s emerging, or optimising for the one that’s already leaving.
The opportunity to reframe the Advertising industry for the better is with us now. Embracing the Agentocene is more than bringing in new technology; it requires an acknowledgement that we can improve what we have created to drive a more equitable ecosystem for advertisers and media owners alike, and to ensure that value and productivity are future-proofed.