What is the future of programmatic advertising?

The structural future of programmatic advertising is two agents transacting: a buyer agent and a publisher agent, with a neutral marketplace governing the deal record between them. The intermediary layer, currently populated by exchanges, SSPs and data aggregators, will compress but not disappear. The timeline depends on the maturity of deal governance infrastructure, not on agent capability, which is already sufficient. By 2028, agent-to-agent deal negotiation will be standard practice at the major agency groups.

The structural future of programmatic advertising is two agents transacting: a buyer agent and a publisher agent, with a neutral marketplace governing the deal record between them. This is not a speculative scenario. The components are in production. What remains is standardisation of the governance infrastructure that makes agent-to-agent transactions verifiable and disputable through machine-readable means.

At a glance

Agent-negotiated deal: A deal committed to by two AI agents without per-transaction human intervention, governed by a bilateral deal record that both parties can query.

Deal governance: The infrastructure layer that records, monitors, and enforces what both agents committed to — price, volume, publisher environment, brand safety — over the life of a campaign.

AI buying agent: A software system that executes deal negotiations within a human-defined mandate, replacing the manual deal management a human trading team would otherwise perform.

AAMP: Agentic Advertising Management Protocols — the IAB Tech Lab standard defining machine-readable deal terms that both buying and publishing agents can parse without human interpretation.

What does the supply chain look like when agents are the primary transactors?

The current supply chain runs from brand to agency to trading desk to DSP to exchange to SSP to publisher, with data and measurement platforms at multiple points. Each layer extracts a toll in the form of fees, latency and information asymmetry. Estimates of the fraction of a programmatic dollar that reaches working media have consistently come in between 50 and 60 cents, with the remainder absorbed by intermediary technology and fees, according to research published by PwC for ISBA in 2020 and updated analyses in subsequent years.

Agent-to-agent transactions compress this chain. A buyer agent, operating within the buyer's mandate, can negotiate directly with a publisher agent without routing through multiple intermediary platforms. The deal record lives in a neutral marketplace that both sides access. The functions of the current intermediary layer: bid clearing, deal management, data enrichment and measurement, are unbundled. Some are absorbed into the agents themselves. Others persist as specialist services that agents call on a per-transaction basis.

Which parts of the current infrastructure survive?

Measurement and verification survive. An agent transacting at speed still needs independent confirmation that impressions delivered matched what was committed and were seen by real people. Third-party measurement is more necessary, not less, in a high-volume agent environment, because human review of delivery is no longer a practical option.

Data infrastructure survives in transformed form. Agents need audience signals to make targeting decisions. First-party data, contextual signals and privacy-compliant identity resolution remain inputs. What changes is how those inputs are accessed: agents query data services rather than human planners integrating data into manual plans.

Ad serving survives. Agents negotiate and commit deals, but the physical delivery of creative to inventory still requires serving infrastructure. That layer does not change structurally.

What are the conditions for this future to arrive?

Three conditions determine the pace of transition. First: agent capability sufficient to handle deal complexity. This condition is largely met. Current large language model-based agents can parse deal terms, evaluate inventory against targeting constraints and negotiate pricing within defined parameters.

Second: governance infrastructure sufficient to make agent-negotiated deals verifiable. This condition is partially met. The AAMP standard provides a protocol framework. What is missing is widespread adoption across buy-side and sell-side platforms. Until deal records are bilaterally accessible in a standardised format, agents that disagree about what was agreed cannot resolve disputes programmatically.

Third: regulatory and contractual frameworks that assign accountability when agents err. This condition is least developed. Advertisers are accountable for what their agents do, but the frameworks for specifying that accountability in contracts and managing it through insurance or indemnity are still being worked out. The Advertising Standards Authority's 2025 consultation on agentic advertising liability is one indicator that regulators are engaging with the question.

How should buyers prepare?

Buyers that want to participate in agent-to-agent advertising need to make three decisions now. What policy framework will govern agent behaviour: which decisions are agent-autonomous and which require human approval? What governance infrastructure will they require from marketplace providers: what does an acceptable deal record look like? And how will they measure agent performance: what is the evaluation framework when agents are making decisions that humans used to make?

These are not technology questions. They are governance questions, and they should be answered before agent deployment, not after the first dispute.

Frequently asked questions

What are AI agents in advertising?

AI agents in advertising are software systems that take autonomous actions within the ad ecosystem — buying, negotiating, optimising — within mandates set by humans, without requiring human approval for each individual transaction.

What is an AI-powered advertising marketplace?

An AI-powered advertising marketplace is a platform that connects AI buying agents to publisher agents, facilitates deal negotiation in machine-readable formats, and maintains bilateral deal records that both sides can query and verify.

How is artificial intelligence changing programmatic advertising?

Artificial intelligence is shifting programmatic advertising from auction-time automation toward proactive deal negotiation: AI agents initiate, negotiate, and commit to deals within human-defined mandates, compressing cycles from weeks to minutes.