Magnite, a company that helps websites and streaming apps sell video ads, worked with a French ad-buying agency called Amnet to let a computer program handle most of the setup work for a car brand's streaming TV campaign. The test matters because it shows what happens when software, instead of a person, picks the ad slots and settings, and Magnite says the campaign got built about 70 percent faster with a high rate of viewers watching to the end. Media buyers and planners now have one more real example, rather than a hypothetical one, of what handing tasks to an AI agent actually looks like in practice.

Magnite this week announced the launch of what it describes as its first agentic campaign in the EMEA region, carried out in partnership with the French programmatic trading desk Amnet France. The announcement, dated September 8, 2026, and published from Paris, credits an approximate 70 percent reduction in campaign setup time and a video view-through rate of 95 to a connected television buy built for an unnamed automotive manufacturer.

Magnite (NASDAQ: MGNI) describes itself as the world's largest independent sell-side advertising company. According to Magnite, the campaign used natural language prompts to let Amnet's team direct a buyer agent that built and activated a video campaign through Magnite's ClearLine platform. The company frames the exercise as evidence that agentic tools can shift media buying teams away from manual configuration work and toward strategic decisions.

What Magnite and Amnet built

According to the announcement, Amnet used natural language instructions to operate Magnite's buyer agent, which in turn built and activated a video campaign through ClearLine, Magnite's curation and activation platform, on behalf of a car manufacturer that the release does not name. The mechanism connecting the two sides is what Magnite calls Orchestration: a system in which the Magnite buyer agent communicates directly with a corresponding Magnite seller agent to identify and activate premium connected television supply that matches the campaign's stated objectives.

That structure removes several steps that a human trading team would normally perform by hand. Rather than manually configuring campaign settings, identifying individual publishers, and constructing deal structures line by line, according to Magnite, the Amnet team directed those tasks through the agent interface and redirected the freed time toward strategy work.

The stated results are twofold. Magnite reports an approximate 70 percent reduction in campaign setup time compared with the manual process the Amnet team would otherwise have followed. The company also cites a video view-through rate of 95, a metric describing what share of a video advertisement's viewers continued watching after the initial exposure point. Magnite additionally states that the agent surfaced inventory and optimization opportunities that a traditional manual workflow might not have identified, which the company frames as improving the quality of campaign decisions rather than only their speed.

The quoted claims

Barbara Thuillier-Romeri, Ad-Tech Manager at Amnet France, is quoted in the announcement describing the motivation behind the test. "As AI continues to mature, we wanted to understand how agentic technology could deliver practical value and complement the way we operate today," she said, according to Magnite. She added that working with Magnite "gave us the opportunity to evaluate how their buyer agent could enhance the way our teams execute campaigns and drive stronger outcomes," and said Amnet intends to keep using more of the product's capabilities going forward.

Edouard Schmidt, Commercial Director for France at Magnite, offered a companion quote framing the result from the supply side. "AI is only as valuable as the inventory and data it can access," he said, according to Magnite. He argued that because the buyer agent sits directly inside the buying workflow, it can surface optimization opportunities as they emerge and let buyers act on them faster, which he said improves both operational efficiency and campaign performance. Schmidt said the results with Amnet reinforce the value of connecting intelligence directly to execution and that Magnite intends to build on the collaboration with more clients across the EMEA region.

Context: a busy year for agentic advertising claims

The Amnet France test lands inside a period of intense activity around AI-driven media buying, much of it tracked in detail by PPC Land. Magnite's own agentic push did not begin with this announcement. The company built a seller agent directly into its SpringServe platform and ran its first agentic advertising test in December 2025, an integration PPC Land reported supported the Advertising Context Protocol, an open standard allowing AI agents on both the buy and sell sides to exchange campaign requirements and match inventory. Scope3 served as the buyer agent in that initial test, executing on behalf of MiQ across LG and Warner Bros. Discovery inventory, a detail Magnite later confirmed on an earnings call, as PPC Land reported.

By April 27, 2026, Magnite had broadened its buyer agent rollout considerably. PPC Land reported that the company introduced AI-assisted mediation features inside SpringServe alongside early tests of the buyer agent with Kepler and MiQ, with access to Disney Advertising inventory, and named Spectrum Reach as an additional participant. Magnite chief executive Michael Barrett has since offered public estimates of how large the category might become. In comments reported by PPC Land in an August 2026 article, Barrett capped his 2027 forecast for agentic advertising spend at roughly 700 million dollars and compared the infrastructure supporting automated transactions to the clearing function behind a stock exchange, describing a model in which one buyer transacts with one seller rather than one party reaching many.

Barrett has also used his public remarks to push back against the idea that AI agents will displace supply-side infrastructure. Speaking on an earnings call covered by PPC Land in March 2026, he argued that AI agents would "make it work better," reducing menial tasks for traders, planners, and operations staff while directing more budget toward working media rather than replacing the platforms that route those budgets.

A crowded field of comparable claims

Magnite is far from alone in publishing performance figures for early agentic campaigns, and the density of similar announcements complicates any single claim's standing as a category first. PPC Land's agentic AI explainer notes that June 11, 2026, alone produced four separate agentic product launches from Mediaocean, Magnite, Teads, and a Walmart Connect and Google DV360 partnership. PubMatic has been especially active in publicizing agentic results through its AgenticOS platform. According to a PubMatic earnings summary covered by PPC Land on August 7, 2026, the company's fact sheet claimed that Amnet delivered a 25.7 percent uplift in video completion in what PubMatic described as France's first agentic campaign, a separate collaboration involving the French livestock trade body Interbev rather than the automotive advertiser named in Magnite's release. Industry coverage from IAB Europe, published around July 20, 2026, described that PubMatic-Amnet campaign as cutting launch time by 80 percent while lifting video completion rate by 25.7 percent, figures that sit alongside but do not match the setup-time and view-through figures in Magnite's announcement.

That overlap illustrates a broader pattern rather than a contradiction specific to Magnite. Amnet, as a trading desk operating across multiple supply-side partners, appears to have run separate agentic pilots with at least two independent sell-side platforms during 2026, each generating its own performance claims and each described by its respective technology partner in terms suggesting a notable milestone. Readers evaluating any single agentic campaign announcement, including this one, are working from vendor-supplied figures that lack third-party verification, a limitation that applies across the sector and not solely to Magnite's release.

PPC Land's ongoing tracking of the category has also surfaced skepticism about whether agentic tools are living up to the efficiency claims attached to them. The agentic AI explainer cites a DataBeat report from June 22, 2026, finding a 13.4 percent cost-per-thousand-impressions advantage for conventional buyers over AI agents in at least one measured comparison, a data point that runs counter to the general narrative of automated efficiency gains that vendor announcements, including Magnite's, tend to emphasize.

The technical mechanism behind the claim

Magnite's Orchestration system, the component connecting buyer and seller agents in this campaign, has been previously described in PPC Land's coverage of the company's broader agentic strategy. An analysis published in June 2026 detailed how the Magnite Seller Agent gives publishers two capabilities: packaging inventory and audiences with flexible pricing and targeting controls, and making that inventory discoverable and purchasable by buyer agents. The described outcome, according to that analysis, is agent-to-agent transactions in which automated systems on both sides of a deal negotiate and complete buys without a human matching supply to demand at each step.

The same analysis noted that buyers who have already built proprietary agentic infrastructure can connect it into Orchestration rather than adopting Magnite's own buyer tooling, a design choice the report characterized as signaling an intent to position the product as neutral infrastructure rather than a closed system. For buyers who do use Magnite's native tooling, as Amnet did in this campaign, the Buyer Agent generates media plans from campaign inputs, surfaces available supply and audience opportunities, and handles creative generation and campaign launch across formats including CTV home screens and audio.

This architecture sits within a wider standardization effort playing out across the advertising technology industry through 2025 and 2026. The Ad Context Protocol, built on Anthropic's Model Context Protocol, launched on October 15, 2025, with six founding members including Scope3, Yahoo, PubMatic, Swivel, Triton Digital, and Optable, and has since drawn additional participants and continued technical development. Magnite's own seller agent, built into SpringServe the following January, was positioned as compatible with that broader push toward agent-to-agent transaction standards, though the ClearLine-based Orchestration system used in the Amnet France campaign appears to represent a parallel, Magnite-specific implementation rather than a direct AdCP integration, based on the available public descriptions of both systems.

About the companies

Magnite describes itself in its announcement as the world's largest independent sell-side advertising company, providing technology that publishers use to monetize content across CTV, online video, display, and audio formats. The company states that its platform is used by agencies and brands to access brand-safe, high-quality ad inventory and to execute billions of advertising transactions each month, with offices spanning North America, EMEA, LATAM, and APAC.

Amnet France, according to the announcement, was founded in 2012 and operates as a programmatic trading desk serving more than 200 advertisers in France and internationally. The company states that it addresses a range of digital marketing challenges spanning brand building, performance marketing, drive-to-store strategies, audience creation, and data-driven targeting, and that it delivers programmatic campaigns across display, video, social, audio, connected TV, and digital out-of-home channels.

Why this matters for marketers

The figures in Magnite's announcement, a 70 percent reduction in setup time and a 95 view-through rate, arrive without independent verification and without disclosure of campaign budget, flight dates, or the specific automotive advertiser involved. That is consistent with how most agentic advertising case studies have been presented across the industry so far: as single-campaign proof points published by the technology vendor rather than as data drawn from audited or peer-reviewed measurement. Marketers assessing whether to test similar agentic tools with their own trading desks have, at this stage, a growing library of vendor-reported figures to compare, but comparatively little standardized methodology for interpreting what those figures mean across different platforms, campaign types, and advertiser categories.

The recurring appearance of Amnet as a named partner across at least two separate agentic campaign announcements, one with Magnite and one previously with PubMatic, according to PubMatic's own materials, underscores how trading desks are positioned at the center of this experimentation. Agencies and their trading arms are running parallel pilots across competing sell-side platforms, which gives holding companies leverage to compare results but also means that headline claims of a market "first" from any single technology partner should be read against the broader pattern of overlapping activity rather than taken as evidence of exclusivity.

For publishers and CTV supply partners, the mechanics described in Magnite's release, particularly the direct communication between buyer and seller agents through Orchestration, point toward a buying environment where campaign setup speed and inventory discovery increasingly depend on how well a platform's own agent tooling is adopted by trading desks, rather than solely on traditional relationship-based sales. Whether that shift proves durable, or whether cost and performance outcomes hold up against the kind of scrutiny DataBeat's June 2026 findings suggest is warranted, remains an open question that will likely require more campaigns, across more advertisers and more sell-side platforms, before clear patterns emerge.

Timeline

  • October 15, 2025: The Ad Context Protocol launches with six founding members including Scope3, Yahoo, and PubMatic
  • December 2025: Magnite runs its first agentic advertising test in SpringServe, with Scope3 as buyer agent on behalf of MiQ
  • January 6, 2026: Magnite confirms it built seller agent capabilities into SpringServe to support AdCP
  • February 25, 2026: Magnite reports fourth-quarter 2025 earnings, disclosing details of its first agent-to-agent CTV campaign
  • April 27, 2026: Magnite expands its buyer agent and SpringServe mediation tools with Disney Advertising, Spectrum Reach, Kepler, and MiQ
  • June 11, 2026: Mediaocean, Magnite, Teads, and a Walmart Connect and Google DV360 partnership announce agentic products on the same day
  • Around July 20, 2026: IAB Europe publicizes a PubMatic and Amnet campaign for Interbev delivering a 25.7 percent video completion uplift, described as France's first agentic campaign
  • August 2, 2026: Magnite chief executive Michael Barrett caps his 2027 agentic ad spend forecast near 700 million dollars
  • August 7, 2026: PubMatic's fact sheet, reported by PPC Land, credits Amnet with a 25.7 percent video completion uplift in a separate agentic campaign
  • September 8, 2026: Magnite announces its first agentic campaign in EMEA with Amnet France, citing a 70 percent setup time reduction and a 95 view-through rate

Summary

Who: Magnite, the independent sell-side advertising company, and Amnet France, a Paris-based programmatic trading desk founded in 2012.

What: Magnite announced the launch of its first agentic advertising campaign in the EMEA region, run in collaboration with Amnet France for an unnamed automotive manufacturer, using Magnite's buyer agent, seller agent, and Orchestration system to build and activate a connected television video campaign through the ClearLine platform.

When: The announcement was published on September 8, 2026, and is dated from that day; the underlying campaign's specific flight dates were not disclosed.

Where: The announcement originated in Paris, France, and covers Amnet France's operations within the broader EMEA region that Magnite serves.

Why: The campaign is presented by Magnite as evidence that agentic buying tools can meaningfully reduce manual campaign setup work, citing an approximate 70 percent reduction in setup time and a 95 view-through rate, at a moment when multiple sell-side platforms, including PubMatic in a separate campaign also involving Amnet, are publishing competing performance claims for early agentic advertising deployments across Europe.