How does agent-to-agent negotiation reduce the number of intermediaries in ad buying?

Agent-to-agent negotiation reduces intermediaries in ad buying by enabling a buyer agent to negotiate directly with a publisher agent, eliminating the role of intermediary platforms whose primary function was to translate between the human-operated systems on each side. The exchange, SSP and trading desk layers exist partly because human buyers cannot directly manage thousands of publisher relationships. Agents can. The functions that survive are those that provide genuine value: clearing, governance, measurement and data services.

Agent-to-agent negotiation reduces intermediaries by making direct buyer-to-publisher deal-making technically practical at scale. The current intermediary layer exists primarily because human buyers cannot maintain direct operational relationships with the number of publishers required to build a meaningful programmatic strategy. Each intermediary layer: trading desk, DSP, exchange, SSP, solves part of that operational problem for humans. Agents dissolve most of it.

At a glance

Supply path: The chain of technology platforms and intermediaries through which an ad impression passes from publisher to buyer — SSP, exchange, DSP, and any resellers in between.

Supply path optimisation (SPO): The practice of reducing the number of hops in the supply path to lower fees, improve transparency, and reduce the latency between buyer intent and publisher delivery.

Bilateral deal record: A deal record held by a neutral marketplace and accessible to both the buying agent and publishing agent — replacing the separate deal records each intermediary currently holds.

Direct deal: A deal negotiated between a buyer and publisher without passing through open auction intermediaries, typically offering more control over inventory quality and pricing terms.

Why are there so many intermediaries in programmatic advertising today?

The intermediary stack in programmatic advertising developed around human operational constraints. A brand cannot directly negotiate with 10,000 publishers. It needs a trading desk to manage the strategy. The trading desk cannot integrate directly with every publisher technology. It needs a DSP. The DSP cannot maintain direct deals with every publisher. It uses exchanges. Exchanges aggregate publisher inventory through SSPs. SSPs represent publisher inventory to exchanges.

Each layer provides a translation service: between human and machine, between buyer scale and publisher fragmentation, between different technical systems. Every layer takes a toll. PwC's 2020 audit of UK programmatic supply chains for ISBA found that 51 pence in every pound reached working media, with the remaining 49 pence absorbed by intermediary technology and fees. A 2023 follow-up found marginal improvement.

Which intermediary functions does agent-to-agent negotiation eliminate?

When a buyer agent can negotiate directly with a publisher agent, the translation functions those intermediaries performed become unnecessary. A buyer agent with a defined mandate does not need a human trading desk to translate strategy into execution. A buying agent implementing A2A and AAMP does not need a DSP to manage the integration with publisher systems.

The exchange's auction-clearing function is partially displaced. Where buyer and publisher agents negotiate deals in advance, those deals do not require auction clearing. Auction infrastructure survives for open exchange inventory where no pre-agreed deal exists, but the proportion of premium inventory traded through negotiated deals rather than open auctions may increase as agent-to-agent deal-making matures.

Which intermediary functions persist?

Measurement and verification persist. An agent transacting at volume needs independent confirmation that what was delivered matched what was committed. Third-party measurement providers: DoubleVerify, IAS, MOAT and equivalents, perform a function that agents cannot perform for themselves. They remain necessary regardless of how the deal was negotiated.

Data services persist. Agents need audience signals to make targeting decisions. First-party data platforms, contextual intelligence providers and privacy-compliant identity services all remain relevant. What changes is the consumption model: agents query these services programmatically rather than human planners integrating them into manual plans.

Neutral marketplace governance persists and in fact becomes more important. When agents are transacting directly at scale, the entity that holds the deal record and governs disputes must be genuinely neutral. An intermediary whose economic interest is aligned with either buyer or publisher cannot reliably perform this function.

What does the compressed supply chain look like?

In an agent-to-agent deal negotiation model, the supply chain has four layers: buyer agent, neutral marketplace (deal record and governance), publisher agent, and ad server (delivery). The functions that currently spread across five or six intermediary layers compress into this structure.

The marketplace layer carries more structural weight than exchanges currently carry. It is not a clearing mechanism but a governance infrastructure: it holds what was agreed, provides both sides access to the deal record, and handles dispute escalation when delivery diverges from commitment.

How much of the intermediary fee is recovered?

The fee recovery from supply chain compression in agent-to-agent buying depends on what proportion of a buyer's spend moves to negotiated deals versus open auction. A buyer that moves a significant proportion of premium video and CTV spend to agent-negotiated direct deals removes exchange and SSP fees from those transactions. On open exchange inventory, the fee structure is less affected.

Industry estimates, including modelling published by the Chartered Institute of Procurement and Supply's 2025 media transparency report, suggest that a buyer executing the majority of premium deals through direct agent negotiation could recover 15 to 25 percentage points of working media efficiency compared to the current programmatic stack. These estimates assume mature marketplace infrastructure and standardised protocol adoption, neither of which is universal in 2026.

Frequently asked questions

What protocols let advertising agents transact directly with publisher agents?

The primary standards are A2A for agent communication and AAMP for deal term structure. A2A handles the messaging layer; AAMP defines what a machine-readable deal record must contain. Agentic marketplace platforms implementing both allow direct agent-to-agent deal negotiation.

How do AI shopping and media-buying agents negotiate prices with publishers?

A buying agent sends a structured deal enquiry specifying price range, inventory requirements, and constraints. A publisher agent evaluates available inventory and responds. Both agents iterate until they reach agreed terms and commit, at which point a bilateral deal record is written.

What is a machine-readable ad deal?

A machine-readable ad deal is a deal record structured as data fields — price, volume, publisher environment, brand safety constraints, delivery period — that both the buying agent and publishing agent can parse and verify without human interpretation.