What is Seller Agent?
A seller agent is software that acts for a publisher, media owner or their representative, describing available inventory in a machine-readable way so authorized AI buying agents can discover it, evaluate it and request deals.
Also known as: sell-side agent, selling agent
How it works
A seller agent sits between a seller's inventory and the AI agents that buy media. Its main job is to translate what a publisher has (sites, apps, CTV channels, formats, audiences, prices, deal types) into structured information that another piece of software can query.
A typical flow looks like this:
- Discovery. A buyer agent asks what is available, often with filters such as geography, device, content category or ad format.
- Evaluation. The seller agent returns matching inventory with the details the buyer needs to compare options, such as pricing or availability.
- Deal request. If the buyer chooses something, the seller agent can start deal creation, often returning a Deal ID that the buyer uses to activate the deal in its buying platform.
Seller agents usually connect through a standard protocol so that each buyer does not need a custom integration. The Model Context Protocol is one such open standard. It lets a server describe the tools it offers so an AI application can discover and call them, and its specification asks hosts to get explicit user consent before invoking a tool (MCP specification). Industry bodies are exploring the same pattern: IAB Tech Lab's AAMP initiative states that "the Agentic ecosystem will start with a Buyer Agent and Seller Agent talking to each other" and publishes reference SDKs for both (IAB Tech Lab, AAMP).
Why it matters
As more buying work moves to AI agents, inventory that cannot be read by machines risks being overlooked. A seller agent is how a publisher stays visible in that new buying path. It also cuts repetitive work: questions that once needed an email thread and a spreadsheet can be answered in a structured way.
Control is the other reason. A seller agent lets the seller decide what is exposed and to whom, rather than having agents scrape public pages and guess. It does not replace supply chain checks such as ads.txt and sellers.json, which still show who is authorized to sell.
Example
Illustrative example. A niche-audience cooking publisher works with a representation partner. The partner's seller agent describes the publisher's web video and native placements by category, geography and deal type. A buyer agent working for a kitchenware brand searches for food content in North America, finds the placements, and requests a private deal. The brand's trader reviews the Deal ID and activates it in the DSP the agency already uses.
IncrementX perspective
IncrementX Seller Agent is an MCP-compatible interface for the inventory IncrementX represents. Authorized AI buying agents can discover inventory by site or app, IAB category, geography, device and environment, ad format, ad size, inventory type, pricing, deal type and availability. They can evaluate and match it, initiate deal creation and receive Deal IDs used to activate through supported buying platforms.
Access is by request and is not public. For publishers, nothing changes in how they work with IncrementX on media representation; Seller Agent is an added route that makes their represented inventory reachable by agentic buyers alongside direct and programmatic demand.