Published: 17 September 2026

TL;DR: Connected television inventory has long been traded through direct insertion orders or programmatic guaranteed arrangements that require significant human involvement at every stage. Agentic buying changes the process, not the principles. A buy-side agent carrying a broadcaster-specific mandate can negotiate directly with a broadcaster's sell-side agent, agree bilateral terms, and write a shared deal record before a single impression is served. The human sets the mandate. The agents do the negotiation. The deal record belongs to both parties.

Connected television inventory sits at the premium end of the programmatic market. High CPMs, limited supply, audience quality that advertisers pay for. Traditionally, that premium has come with a corresponding overhead: direct insertion orders negotiated between human buyers and broadcaster sales teams, or programmatic guaranteed deals set up manually through SSP and DSP platforms, with post-campaign reconciliation that compares two separate records held by two separate parties.

Agentic buying does not change what CTV inventory is or what makes it valuable. It changes who conducts the negotiation, how long that negotiation takes, and where the deal record lives.

How the Buy-Side Agent Carries the Mandate

A buy-side agent entering a CTV negotiation does not act on open instructions. It operates under a mandate: a human-approved governance document that defines the parameters within which the agent can agree a deal.

For CTV inventory, a well-written mandate covers the content categories the brand will appear against, the CPM ceiling by category, minimum completion rate requirements, daypart restrictions, and the budget thresholds above which the agent must escalate to a human before agreeing. The mandate is set before the agent begins any negotiation. The agent cannot agree terms outside the mandate.

This is the principle of earned autonomy: the agent operates with real authority within defined parameters, and escalates anything that falls outside them. The human does not supervise each negotiation. The human sets the parameters, and the mandate enforces them.

How the Broadcaster's Sell-Side Agent Evaluates Proposals

The broadcaster's sell-side agent holds a corresponding set of parameters: CPM floors by content category, content adjacency requirements (which advertiser verticals can appear against which programming), audience data conditions, and minimum deal sizes.

When the buy-side agent submits a proposal, the sell-side agent evaluates it against these parameters. If the proposal meets the floor and the content adjacency requirements, the sell-side agent can accept. If it falls short, it can counter-propose or decline. The negotiation follows a defined sequence of message types: proposal, counter-proposal, acceptance, rejection.

This is the AdCP (AdCP, published by AgenticAdvertising.org) in operation. Both agents speak the same protocol. The negotiation does not require a human intermediary to translate between platforms or systems.

What Changes Compared to Conventional CTV Buying

Three things change when you move from conventional CTV buying to agent-to-agent negotiation.

Speed. A conventional direct IO negotiation involves email, calls, and manual deal configuration. An agent-to-agent negotiation runs in the time it takes two compliant systems to exchange messages. Multiple broadcasters can be negotiated with simultaneously, because the buy-side agent is not limited by the number of conversations a human buyer can have in parallel.

The deal record. In programmatic guaranteed, the buyer's DSP holds one record and the publisher's SSP holds another. Reconciliation post-campaign means comparing the two. In agent-to-agent negotiation, both parties hold the same DealSheet: a bilateral record written at negotiation time, before any impression is served. There is no reconciliation problem because there is no divergence between the two records.

Audit access. Because the DealSheet is bilateral, both the buyer and the broadcaster can retrieve it independently post-campaign. Neither party is dependent on the other's willingness to share data to verify what was agreed.

What Stays the Same

Three things do not change.

Human mandate approval. The agent does not set its own parameters. A human at the buying organisation approves the mandate before the agent begins any negotiation. The CPM ceiling, the content category restrictions, the completion rate requirements: all of these are human decisions expressed in a documented mandate.

Broadcaster control over CPM floors. The broadcaster's sell-side agent enforces the floors the broadcaster has defined. A buy-side agent cannot negotiate the broadcaster below its stated floor. The broadcaster's commercial interests are protected by the sell-side mandate, not by a human monitoring every incoming proposal.

Documentation before delivery. The DealSheet is written at agreement time. No impressions are served until the deal record exists. This is the same principle as a signed IO: the agreement comes first.

Why Scale Matters

The practical advantage of agent-to-agent CTV buying is not simply that it is faster than human negotiation for a single deal. It is that the same buy-side agent can run parallel negotiations with multiple broadcasters simultaneously, all within the same mandate parameters.

A buyer who wants to reach audiences across several broadcasters' CTV inventory currently needs a human team to manage multiple direct relationships. With agent-to-agent buying, the mandate defines the parameters once, and the agent applies them consistently across every negotiation. The mandate becomes the scalable unit of buying governance.

That consistency is also what makes the audit straightforward. Every deal the agent agreed was within the mandate. Every deal it declined was outside the mandate. The decision log records both.