Published: 17 September 2026

TL;DR: Programmatic guaranteed and agent-to-agent CTV buying both aim to transact premium inventory at pre-agreed prices. The structural difference is in who holds the deal record, how that record is written, and how reconciliation works post-campaign. In programmatic guaranteed, buyer and publisher hold separate records and reconcile after the fact. In agent-to-agent buying, both parties hold the same bilateral DealSheet written at negotiation time. That single structural difference has consequences for every downstream process.

Programmatic guaranteed was designed to bring efficiency to direct-sold CTV inventory. Instead of a fully manual insertion order process, programmatic guaranteed allows buyers and publishers to set up fixed-price, fixed-volume deals through DSP and SSP platforms. The deal is still negotiated by humans, configured manually on both sides, and documented separately by each party's platform. What changes is the activation layer: the deal is trafficked programmatically rather than through manual ad server line items.

Agent-to-agent CTV buying changes the negotiation layer, the deal record structure, and the reconciliation model. It is not a version of programmatic guaranteed with faster deal setup. It is a different structural approach to bilateral CTV deal-making.

How Programmatic Guaranteed Works Today

In a typical programmatic guaranteed arrangement, a broadcaster's sales team and a buyer's team agree terms over email or calls. Those terms are then manually configured: the buyer sets up the deal in their DSP, the publisher sets up the corresponding deal in their SSP, and the two platforms communicate via deal ID when a matching impression is available.

The buyer holds a record of the deal in their DSP. The publisher holds a record in their SSP. These records are created independently and are not inherently synchronised. Post-campaign, the buyer and publisher compare delivery data and agreed terms. If the records diverge, the discrepancy needs to be resolved between the two parties.

Deal setup requires human configuration on both sides. Each party's record is separate. Reconciliation is a comparison exercise between two independent datasets.

How Agent-to-Agent CTV Buying Works

In agent-to-agent CTV buying, the negotiation is conducted between a buy-side agent operating under a mandate and a sell-side agent operating under the broadcaster's deal parameters. The agents communicate using AdCP (the IAB Tech Lab Agent Communication Protocol), which defines the message format, the proposal schema, and the negotiation sequence.

When the agents reach agreement, the agreed terms are written to a DealSheet: a bilateral record that both parties hold simultaneously. The DealSheet is written at negotiation time, before any impression is served. It is not created by one party and shared with the other. It is the shared output of the negotiation itself.

There is no separate DSP record and SSP record that need to be reconciled. There is one agreed record, held by both parties.

The Structural Difference: Who Holds the Record

This is the core distinction. In programmatic guaranteed, the deal record is held separately by each party's platform. In agent-to-agent buying, the deal record is bilateral.

What this means for reconciliation. In programmatic guaranteed, post-campaign reconciliation compares two records that may have diverged through configuration errors, delivery discrepancies, or system timing differences. In agent-to-agent buying, there is a single shared reference. Delivery comparison works from a common baseline.

What this means for audit. In programmatic guaranteed, a buyer auditing a campaign retrieves their DSP record. If that record is incomplete or incorrect, there is no independent reference. In agent-to-agent buying, the buyer can retrieve the DealSheet independently of the publisher. The record does not belong to either party's platform. It belongs to both parties equally.

What this means for dispute resolution. When programmatic guaranteed delivery disputes arise, the two parties compare their separate records. When an agent-to-agent deal dispute arises, both parties refer to the same record.

The Operational Difference: Parallel Negotiation

Beyond the deal record structure, there is a practical operational difference. Programmatic guaranteed requires a human to negotiate and configure each deal. A buyer who wants programmatic guaranteed arrangements with five broadcasters needs five separate negotiation processes run by a human team.

An agent-to-agent buy-side agent can run parallel negotiations with multiple broadcasters simultaneously, within the same mandate parameters. The agent applies the mandate consistently across every negotiation. A buyer who wants deals with five broadcasters does not need five separate human-managed processes. The mandate sets the parameters once. The agent applies them across all five negotiations.

This is not a marginal efficiency gain. It changes the scale at which bilateral CTV deal negotiation is operationally viable for a buying team.

What Programmatic Guaranteed Still Does Well

Programmatic guaranteed is not made redundant by agent-to-agent buying. It remains appropriate for buyers who prefer to manage deals through their DSP, who have existing programmatic guaranteed relationships with broadcasters, or whose deals involve terms that require human review on both sides.

The choice between the two models depends on the buyer's priorities. If bilateral deal record integrity and parallel negotiation scale are the priorities, agent-to-agent buying addresses both. If the priority is managing all inventory through a single DSP interface, programmatic guaranteed remains the appropriate structure.

What agent-to-agent buying provides is a structural alternative. The bilateral DealSheet, the mandate-governed negotiation, and the parallel negotiation capability are properties of the structural model, not features that can be added to programmatic guaranteed through configuration.