Key takeaway: For teams building AI marketing operations, Pinterest’s announcement addresses where inventory can be activated, while the Kovva interview a.
Key Takeaway
For teams building AI marketing operations, Pinterest’s announcement addresses where inventory can be activated, while the Kovva interview addresses how buyers can manage work that crosses platforms. Together, the reports support a practical operating rule: broader programmatic access and agentic tooling can reduce friction, but they do not remove human control over budgets, campaign setup, or other hard-to-reverse changes. Pinterest was not reported to have moved all of its ad revenue into programmatic channels, and the reporting does not establish that AI agents can independently manage a complete media plan.
What Pinterest’s programmatic access changes for ad platforms
Pinterest announced general availability of Programmatic Deals in North America, South America, and Asia Pacific. Buyers can transact Pinterest display and video inventory through Amazon Ads, Google Display & Video 360, and Yahoo DSP. The available deal types include private marketplaces, preferred deals, and programmatic guaranteed deals. Magnite is Pinterest’s primary supply-side platform, while Google Ad Manager handles supply in Latin America and Asia Pacific.
The operational change is straightforward: Pinterest inventory can enter buying workflows that agencies and brands already use, rather than being available only through Pinterest’s own advertising tools. Pinterest said its investment in programmatic is intended to make the platform available wherever buyers want to activate media. The report also says Pinterest inventory is part of Magnite Orchestration, a layer where AI agents can find, evaluate, package, and activate ad inventory. That description shows a possible path into agentic media buying; it does not prove that every advertiser is using an agent to buy Pinterest inventory or that an agent can act without account-level controls.
Pinterest described itself as a visual search and shopping experience and said it has 640 million monthly active users and more than 80 billion searches each month, with more than half of those searches being commercial. Those are figures and characterizations attributed to Pinterest. They should not be converted into a guarantee of SEO performance, AI search visibility, or advertising conversion. A buyer still needs to validate audience quality, frequency, supply conditions, measurement, and attribution for the specific deal and DSP.
What cross-platform AI agents can and cannot automate
AdExchanger’s interview with Tanja Mimica, COO and co-founder of Kovva, focuses on the work media buyers perform between systems. Major advertising platforms are building AI to optimize activity inside their own environments, but buyers still pull reports from DSPs, social platforms, and ad servers, reconcile mismatched numbers, quality-assure campaigns, and investigate undelivered line items. Mimica’s central observation is that much of a buyer’s work happens between platforms rather than inside one platform.
Kovva was described as connecting to about 50 tools that buyers already use, including DSPs, search and social platforms, verification partners, Slack, and Teams. Buyers commonly begin with a daily or weekly briefing, compare it with the underlying platforms, ask follow-up questions, and then schedule recurring tasks after the system has demonstrated reliability. The interview says the transition to more complex delegated workflows takes at least three to four weeks. For an advertising operations team, this model could help consolidate checks when a new source such as Pinterest is added to an existing DSP workflow.
The approval boundary is equally important. The interview says buyers remain cautious about high-stakes or difficult-to-undo actions, such as setting up a campaign with a large budget. Kovva’s guardrails are established outside the agents by people. An agent cannot raise its own budget limit or approve its own change, and, at least for now, every “write” action requires human approval. In practical terms, an AI marketing team can begin with read-only reporting, discrepancy checks, delivery alerts, and recommendations. Budget changes, campaign launches, targeting edits, and supply-source changes should remain in a review queue with an accountable human.
How to read Pinterest’s performance evidence
Pinterest shared one Moroccanoil example. In June, the beauty brand added a Pinterest deal as a supply source to an existing Amazon DSP campaign. Pinterest said the addition increased campaign reach by 15%, with 1% overlap between Pinterest’s audience and people the campaign was already targeting. It also said Pinterest click-through rates were 20% higher than Moroccanoil’s average Amazon DSP campaign. These figures describe one campaign and retain the comparison supplied in the report. They are not evidence that every brand or DSP will see the same outcome, nor do they independently prove incremental sales.
An Amazon DSP buyer can use Amazon Marketing Cloud to connect a Pinterest ad with a purchase on Amazon, according to the report. That is a measurement path, not an automatic attribution result. Teams still need to define events, attribution windows, audience overlap, inventory quality, and any control or holdout design. Pinterest also declined to disclose what share of its advertising revenue comes through programmatic channels, saying only that the share is growing. Advertisers were described as treating programmatic as an addition to Pinterest’s native tools.
The appropriate conclusion for operators is therefore narrower than a growth claim. AI agents may help inspect and coordinate buying across platforms, while Pinterest programmatic access adds another source to the media mix. Neither report establishes that the combination improves SEO, Google search ranking, or visibility in AI search answers. Pinterest’s visual-search positioning may be relevant to commercial intent, but it is not evidence of a direct organic-search or AI-search ranking effect.
Method and Limitations
When adding a programmatic source, a team can first document the available DSPs, deal types, regions, supply partners, and account permissions. It can then give an agent read-only tasks such as summarizing the last 24 hours of delivery, comparing platform reports, flagging discrepancies, and checking spend against limits. Any write action should require named approval, a defined budget ceiling, and a recoverable record of the change.
The Moroccanoil figures should remain labeled as a single platform-reported example, including their 15% reach increase and 20% click-through-rate comparison. They should not be generalized to other accounts. The public reporting does not establish Pinterest’s programmatic revenue share, whether The Trade Desk is a launch DSP, or how much real Pinterest buying is being automated by AI agents. It also does not establish an SEO or AI-search ranking benefit. Teams seeking an incremental result should use a defined geographic, audience, or supply-source comparison rather than relying on platform positioning or an agent-generated summary.