AI Agents Are Starting to Buy Media

Advertising decisions help determine what entertainment earns

Advertising can help pay for a creator's next video, a live sports broadcast or a Hollywood series on an ad-supported streaming service. Buyers decide which audiences and viewing environments deserve that money. AI agents are beginning to take on more of the campaign work behind those decisions. How those agents evaluate a campaign matters to the businesses that create and distribute the entertainment.

An agent, in this context, is software that carries out steps toward a goal, with authority set by the person using it. Advertising platforms have automated bids and budget pacing for years. Newer agents extend that automation into setting up campaigns and investigating their performance.

The Trade Desk, a platform advertisers and agencies use to buy digital ads, gives us a concrete example. It describes an assistant called Koa that can take an uploaded media plan, help build a campaign and flag inconsistencies. Its campaign-creation process includes human review and approval. The assistant is in closed beta, and some described capabilities are still being rolled out.

A TikTok video and a premium streaming drama give advertisers different settings in which to reach people. Their buying systems and commercial arrangements also differ. Koa is one example within The Trade Desk; it does not establish a single buying system across those environments.

Less manual setup could let a small team manage more campaigns. Faster analysis could give a buyer time to adjust spending while a campaign is still running. The benefit depends on how much work the software handles correctly and how much checking remains.

That creates a pricing question for agencies: if software takes on more campaign work, what will clients pay the agency to do? Digiday reports that agencies are funding AI through existing technology budgets and testing ways to pass costs to clients. One is experimenting with billing AI usage separately from human labor. A faster campaign setup can still come with a new software bill.

For a streaming service or creator business, the stakes include whether its audience attracts that spending, and at what price. A buyer needs to judge whether an ad suits the surrounding content and whether it produces results. A seller needs evidence that helps justify the price it asks.

My conviction is that dependable measurement and spending controls will become more important to media revenue as buyers delegate decisions to software. The criteria those systems use could influence which content businesses earn the spend. Founders have an opportunity to help buyers assess those differences and act on them.

That last question means identifying sales the advertising actually caused. A sale credited to a campaign may have happened anyway. The calculation also needs to include software fees and the time people spend checking or correcting the work.

Existing buying platforms already have customer relationships and can build their own controls. A separate provider has to earn its place. Consistent spending rules across several platforms, or an independent check on results, could justify another purchase. The proof will be customers paying again after seeing the full bill.

Making transactions and costs easier to check

What the product changes for buyers and media owners

When we look at ad-tech companies now, we ask how their software changes the economics for the advertiser and the media owner. Can a buyer judge what an audience is worth for a particular campaign? Can a streaming service or creator business demonstrate why its content deserves the spend? Who pays for that capability?

I want founders to show how their knowledge of media workflows improves those decisions. A product that helps a buyer evaluate content suitability, or a seller substantiate campaign results, has a specific job to do. The diligence question is whether customers will pay for it repeatedly.

Our September 27 review covered roughly 3 months of pipeline arrivals: about 160 companies, including roughly a dozen with advertising themes and about 10 with agent-workflow themes. These overlapping categories describe pitches, not adoption.

The pitches share a broader ambition: software that manages more of the work involved in creating, distributing and monetizing media. Advertising founders are proposing more control over how that work reaches a customer. Agent founders are proposing ways to coordinate work across tools.

Since we introduced Wide Worlds in our last issue as our most recent addition to our portfolio, the team has been running paid creator marketing campaigns for brands, music labels and entertainment companies. Its platform uses agents to manage hundreds of fan creators making and posting content. Those agents vet creators for brand fit, check submitted posts against the campaign brief and manage budgets across creator reward tiers. Wide Worlds reports that September revenue grew 57% over August and that it has retained 87% of its customers since signing its first customer in June. When a buyer hands execution to software and a large pool of participants, the brief, the budget rules and the record of what was paid become part of the product.

The Wide Worlds example concerns paid creator work, distinct from agents buying ad placements. Both require software to enforce campaign instructions. Taste and cultural timing still require human judgment.

Repeat campaigns will test the economics

Over the next 6 to 18 months, I'll be watching for repeat campaigns with a complete account of their economics.

A buyer increasing its commitment after measuring the full cost would strengthen the case for these products. Persistent corrections, unclear attribution or heavy implementation work would weaken it.

For media owners, I'll also watch whether these systems help them demonstrate the value of their audiences and win repeat advertiser spending. A faster buying process creates a useful business only if the participants have an economic reason to keep using it.