A media buyer needs a buy-side agent first, because a buy-side agent is the only agentic tool that acts on the buyer's behalf. Sell-side agents belong to publishers and other inventory owners: they package inventory, hold price floors and accept or counter offers. The real decision for a buyer is not which side to start with. It is whether the sellers it already trades with can respond to an agent, and whether the terms the two agents agree end up in a record both sides hold.

TL;DR. Buy-side agentic tools act for advertisers and agencies; sell-side agentic tools act for publishers. A media buyer adopts a buy-side agent and checks that its sellers run sell-side agents, rather than choosing between the two. IAB Tech Lab's AAMP framework publishes separate reference implementations of a buyer agent and a seller agent, which reflects how the industry divides the roles. Before a buy-side agent spends money, it needs a written mandate, a list of sellers whose agents it can reach, and an agreed record of each deal.

What is the difference between buy-side and sell-side agentic tools?

The difference is whose interests the agent represents. A buy-side agentic tool works for the party spending the money. It takes a brief, finds inventory that fits, reads proposals, negotiates within a mandate, places the buy, paces spend and reports delivery. Its measure of success is value for the advertiser.

A sell-side agentic tool works for the party holding the inventory. It describes the publisher's products, responds to buyer requests, enforces floors and deal rules, and accepts, counters or declines proposals. Its measure of success is yield for the publisher.

What separates the two is the principal each one answers to, not the model underneath. IAB Tech Lab's AAMP framework, the Agentic Advertising Management Protocols, makes the split explicit by publishing a reference buyer agent (demand-side logic and decision workflows) and a reference seller agent (supply-side coordination and marketplace interaction) as separate components.

Which does a media buyer need first?

A media buyer needs the buy-side agent first, and in most cases only the buy-side agent. A sell-side agent negotiates against the buyer on the publisher's behalf, so running one would mean representing the other side of the deal.

The sell-side question still matters, as a check rather than a purchase: a buy-side agent has little to do if no seller can answer it. Before committing, a buyer can list its most important publishers and ask each one three things: does it run a seller agent, how does it describe its products to agents, and which deal types can that agent agree.

The exception is a business that holds inventory as well as buying it, such as a retailer running its own media network. It sits on both sides and may need both kinds of tool, under separate mandates. For an advertiser, or an agency buying for clients, the order is simple: buy-side agent first, sellers' readiness checked alongside.

What should a buy-side agent be able to do on day one?

A buy-side agent earns its place on four tasks, each testable in a pilot.

Read offers in a common format. IAB Tech Lab's OpenProposal specification, part of AAMP, standardises how publishers define products and packages in RFP responses, so an agent can compare sellers without translating each one.

Work across the deal types the desk already uses. According to IAB Tech Lab, AAMP's support for the Deals API covers programmatic guaranteed, preferred deals, private marketplaces, the open marketplace and linear GRPs.

Stop at its limits. The agent should halt and pass a decision to a named person when a proposal crosses a spend threshold, breaches a price ceiling or touches a term it was told not to trade.

Show what it agreed. Every deal should leave a record of the terms that the seller holds in the same form, so delivery is checked against one version.

Why does the mandate matter more than the tool?

A buy-side agent is only as good as the limits it is given, and closing a deal is not the same as closing a good one. In Concourse Bench v1, a benchmark of AI buying agents negotiating with software sellers, 95 of the 110 contracts signed promised a viewability floor below 70%. The agents completed deals. Where no one had defined an acceptable floor, the floor they accepted was low.

The remedy is a mandate written before the agent goes live. For example, a hypothetical £40,000 display campaign might set a maximum CPM, a minimum viewability floor, excluded content categories, a frequency cap, and a rule that any single commitment above £10,000 waits for a named person to approve it. Each line is a term the agent may not negotiate away, and each gives the buyer something specific to check after delivery.

What does a sell-side agent change for the buyer?

A sell-side agent answers within its own rules and holds its floor the same way every time. That is faster than a sales team replying to an RFP, and less flexible: it will not bend a rule that a person on a sales call might.

Buyers adjust in two ways. Briefs need to be precise, because an agent reads ambiguous terms literally. And every counter-offer needs to be checked against the mandate rather than the last deal, because a seller agent working for yield will propose terms that suit the publisher and may sit outside what the buyer authorised.

How should a buyer judge when to switch on a buy-side agent?

Timing is better judged by signals than by a date. Four observable indicators show whether a buyer's own market is ready: seller coverage (how many priority publishers can respond to an agent at all), format (whether they describe products in a shared format such as OpenProposal), discovery (whether agents on both sides appear in a public directory such as the IAB Tech Lab Agent Registry), and the record (whether each agreed deal produces terms both parties hold).

A buyer who sees all four in place for a meaningful share of its spend has a case for a pilot. A buyer who sees none should ask its sellers again in a few months and use the time to write the mandate.

Where does the shared deal record sit?

Neither agent settles what was agreed. Each keeps its own logs, and in a delivery dispute two sets of logs are two versions of events. The record both sides accept has to sit between the agents rather than inside either one.

That middle position is where Alkimi operates. As a neutral agentic advertising marketplace working within IAB Tech Lab's AAMP framework, it checks that each connecting agent is authorised by the buyer or publisher it represents, and the terms the agents agree are written to a DealSheet that both sides hold, with people keeping approval rights over each deal.

Frequently asked questions

Does a media buyer need a sell-side agent?

In most cases, no. A sell-side agent represents the publisher. A media buyer needs a buy-side agent, and needs its sellers to run sell-side agents that the buy-side agent can reach.

What is a buy-side agentic tool?

Software that acts for an advertiser or agency within a mandate: it finds inventory, reads proposals, negotiates terms, places buys, paces spend and reports delivery, passing decisions outside its limits to a person.

Is a buy-side agent the same as a DSP?

No. A DSP prices individual bid requests in an auction. A buy-side agent compares offers, negotiates deal terms and manages a budget across many transactions.

What should a buyer ask publishers before using a buy-side agent?

Whether they run a seller agent, how they describe their products to agents, which deal types the agent can agree, and whether agreed terms are recorded in a form both parties hold.

Who is accountable for a deal a buy-side agent agrees?

The organisation that deployed the agent. An agent is not a legal entity, so its commitments belong to its principal, which is why the mandate and the deal record matter.

Buy-Side or Sell-Side Agentic Tools: Which Does a Media Buyer Need First?