In October, a consortium of companies announced Ads Context Protocol (AdCP), which promises to standardize how machines, platforms, and AI agents talk to one another across advertising platforms – from connected TV to digital out-of-home.
The protocol arrived with significant industry backing: over 20 leading advertising technology companies, including Yahoo, PubMatic, Magnite, Optable, Scope3, Swivel, and Triton Digital, are involved. A coalition of heavyweights doesn’t form unless the pain is real and the solution credible, open source, with neutral governance.
AdCP’s promise is elegant: standardise how AI agents discover inventory, negotiate deals, and execute campaigns across any platform, eliminating the fragmented API landscape that has plagued advertising for two decades.
But as the industry stands at this inflection point, critical questions emerge: will AdCP deliver genuine transformation, or is it merely a sophisticated repackaging of persistent problems?
Evolution, not extinction, for programmatic buying
AdCP doesn’t kill traditional programmatic – it rescues premium inventory from being forced into a commodity model it was never suited for. Publishers have spent 15 years watching their differentiated audiences get reduced to CPM line items in a race-to-the-bottom open exchange. AdCP gives them a way to expose that value to automated systems without surrendering pricing power.
Programmatic becomes the execution layer that sits beneath strategic agent negotiation. The agents negotiate the deal terms via AdCP; the traditional programmatic infrastructure handles the actual ad delivery and optimisation. This is evolution, not extinction.
Where traditional programmatic dies is in manual campaign setup and platform-by-platform optimisation. That workflow is genuinely obsolete. But the underlying RTB auction mechanics? Those stick around – they just get orchestrated by agents instead of humans.
The rise of Agentic Advertising
If AdCP is the connective tissue, then agentic advertising is the nervous system.
It’s likely we are about 18 months away from the first seven-figure campaign that executes entirely via agent negotiation – and maybe two years away from the first spectacular agent screw up that becomes a cautionary tale at every industry conference for the next decade.
Here’s where agentic advertising gets interesting: truly autonomous doesn’t mean “zero human involvement” – it means humans shift from execution to strategy. You set the parameters: budget thresholds, brand safety guardrails, outcome metrics, and approval requirements for deals above certain thresholds. The agent handles discovery, negotiation, and optimisation within those boundaries.
It’s likely we are about 18 months away from the first seven-figure campaign that executes entirely via agent negotiation – and maybe two years away from the first spectacular agent screw up that becomes a cautionary tale at every industry conference for the next decade.
Should humans ever be fully out of the loop? In tactical execution, yes. But not in strategic decisions.
Here’s a recommended framework:
Agent-autonomous: Media mix optimization, bid management, creative rotation, performance monitoring, budget pacing, audience discovery.
Human-approved with agent recommendation: Major budget shifts, new platform adoption, brand partnership deals, crisis response, anything involving sensitive content adjacency.
Human-only: Brand positioning strategy, legal/compliance decisions, crisis communications, anything involving trade-offs between short-term performance and long-term brand value.
The danger isn’t agents making decisions, it’s us abdicating the responsibility to set proper boundaries. Every time an agent makes a bad decision, it’ll be because a human didn’t define the right constraints or ignored warning signs because the vanity metrics looked good.
Publishers reclaim power
AdCP hands publishers their leverage back. For the first time in 15 years, they can expose premium inventory directly to buyer agents with transparent pricing, contextual data, and packaged audience segments – without getting forced into lowest-common-denominator open exchange dynamics.
Buyers benefit from reduced complexity: one interface instead of eight platform logins. AI companies benefit from a standardised protocol to build on. But those are operational wins. Publishers get something more fundamental: the ability to compete on value instead of just volume.
The Transparency Question
The sceptic in me asks: will walled gardens implement AdCP faithfully, or will they add proprietary extensions that recreate fragmentation?
Here’s what gives me cautious hope: AdCP’s architecture includes audit trails, data provenance requirements, and asynchronous design allowing human oversight. Every agent conversation gets logged. Every decision gets documented. Unlike millisecond RTB auctions, where humans can’t possibly intervene, AdCP workflows operate at human-comprehensible timescales.
But – and this is critical – we need binding commitments: mandatory audit access, immutable transaction logs, standardised reporting across platforms, independent verification mechanisms, and clear accountability when agents make decisions that violate brand safety, waste budgets, or optimise for the wrong outcomes.
AdCP provides infrastructure for transparency. What we do with that infrastructure determines whether it becomes the open accounting ledger that advertising desperately needs, or just a faster, fancier way to obscure where the money actually goes.
By 2030, agentic workflows will make up 70-80% of digital advertising executions. This is structurally inevitable.
My prediction: we’ll get both. Some implementations will be radically transparent because brands demand it and independent publishers see it as a competitive advantage. Others will be “technically compliant,” where the audit logs exist but are deliberately unhelpful.
As media agencies and advertisers, we should not accept whatever dashboards we’re given – it’s vital we demand actual transparency.
It’s likely we are about 18 months away from the first seven-figure campaign that executes entirely via agent negotiation – and maybe two years away from the first spectacular agent screw up that becomes a cautionary tale at every industry conference for the next decade.
Humans top out. Agents don’t. Manually logging into ad platforms will soon be as archaic as faxing insertion orders.
By 2030, agentic workflows will make up 70-80% of digital advertising executions. This is structurally inevitable. It has become impossible to manually manage real-time budget allocation across CTV, social, programmatic, retail media, and emerging channels while accounting for attribution, incrementality, brand safety, and compliance. Humans top out. Agents don’t. Manually logging into ad platforms will soon be as archaic as faxing insertion orders.
My specific prediction: The buzzword phase ends in 2026. The backbone building phase is 2027-2029. The battleground? That’s permanent. Because advertising is where technology, creativity, commerce, regulation, and human psychology collide – and AI just made that collision a hell of a lot more complex.
For the foreseeable future, battles will rage around governance, accountability, transparency, trust and compliance panic. One high-profile failure could set adoption back years.
So buckle up – it’s going to be an interesting ride.
Anil is Managing Partner–Data Strategy and Partnerships at Publicis Media India.
