The Trade Desk Is Letting Advertisers Build Campaigns Through Claude. Programmatic Will Never Be the Same.

Twenty million ad impression opportunities hit The Trade Desk's platform every second. Millions of campaigns. Billions of users on the other side. Each decision window lasts less than 10 milliseconds. And right now, a small group of advertisers is using Anthropic's Claude to make those decisions.

The Trade Desk CEO Jeff Green confirmed at Marketecture Live in New York City on March 11 that the company is running a closed beta allowing select advertisers to create campaigns on its platform using generative AI. The test uses Anthropic's Claude, positioning The Trade Desk as one of the first major demand-side platforms to directly integrate a frontier AI model into campaign creation workflows.

Green was characteristically blunt about the opportunity. "I don't think that there is an industry in the world that is more conducive to AI than programmatic advertising," he told the audience during a panel with Marketecture Media founder Ari Paparo.

He stopped short of sharing specifics about the beta's scope, the number of advertisers involved, or when a broader rollout might happen. What he did share was a strategic vision that could reshape how billions in ad spend get allocated.

Green's argument isn't just corporate positioning. It's mathematically sound.

Programmatic advertising involves real-time decisions at a scale that exceeds human cognitive capacity by several orders of magnitude. A single campaign might evaluate billions of impression opportunities across thousands of audience segments, creative variations, and bidding strategies. Today, campaign managers handle this complexity through a combination of rules-based automation, manual bid adjustments, and platform-specific optimization algorithms. The process works, but it's slow, labor-intensive, and constrained by the number of variables a human can meaningfully manage.

Generative AI changes the input layer. Instead of navigating complex platform interfaces, setting bid parameters, and manually configuring targeting, an advertiser could describe their campaign objectives in natural language and let Claude translate that intent into platform-specific configurations.

That's the same paradigm shift we've seen in software development with AI coding assistants. Developers describe what they want; the AI writes the code. Green is betting that advertisers will describe what they want; Claude will build the campaign.

Green offered limited specifics. Here's what's confirmed from his Marketecture Live appearance:

What wasn't said matters too. Green didn't clarify whether Claude has access to live performance data during campaign creation, whether advertisers can iterate on campaigns through conversational interaction, or how the system handles guardrails around budget allocation and brand safety. Those details will determine whether this is a meaningful product or a polished demo.

The Trade Desk isn't building in a vacuum. Every major ad platform is racing to deploy AI agents, and the competitive dynamics are intensifying weekly. The distinction that matters is between AI-assisted optimization (which every platform already does) and AI-driven creation (which is what The Trade Desk is testing). Optimization means the platform improves performance within parameters humans set. Creation means the AI sets those parameters based on advertiser intent. That's a fundamentally different level of autonomy.

For performance marketing managers, the practical implication is significant. If AI can reliably translate business objectives into campaign configurations, the value of deep platform expertise shifts. Knowing how to navigate The Trade Desk's interface becomes less important than knowing how to clearly articulate campaign strategy. The skill set changes from technical execution to strategic communication.

The most provocative part of Green's appearance wasn't about AI. It was about Amazon.

Green reiterated his view that Amazon will eventually retreat from operating a demand-side platform for the open internet, citing antitrust risk as the primary driver. He pointed to Google's ongoing regulatory battles as a cautionary example, arguing that Amazon's participation in open web ad tech intersects with retail, cloud infrastructure, and streaming video in ways that create enormous legal exposure.

That's a bold prediction from a CEO whose company directly competes with Amazon's DSP. Green has clear financial motivation to talk down Amazon's open web ambitions. But his logic isn't easily dismissed. Amazon's advertising business generated $56.2 billion in 2025, the vast majority from product listing ads within its own ecosystem. The open web DSP business is a fraction of that. If operating it invites antitrust scrutiny that threatens the core retail and cloud businesses, the math doesn't work.

Whether you buy Green's thesis or not, his willingness to state it publicly signals confidence in The Trade Desk's competitive position. That confidence is partly grounded in the AI roadmap he's now telegraphing.

If you're managing programmatic campaigns today, the Claude beta is both exciting and unsettling. Exciting because AI-driven campaign creation could dramatically reduce the time and expertise required to launch sophisticated programmatic strategies. Unsettling because it raises questions about the future of the specialized skills that currently justify performance marketing salaries.

The realistic near-term scenario is augmentation, not replacement. AI will handle the mechanical complexity of campaign configuration while humans focus on strategy, creative direction, and performance interpretation. The Trade Desk's beta likely confirms that approach: advertisers describe intent, Claude builds the framework, humans review and approve.

For agencies, the competitive pressure is acute. If brands can create campaigns through natural language interaction with an AI model, the value proposition of agency media buying shifts toward strategy, creative, and analytics rather than platform management. Agencies that have already been investing in strategic capabilities will see this as validation. Those still selling execution will feel the squeeze.

The cost implications deserve attention too. Green noted during his remarks that AI companies face a serious monetization problem, with massive capex and sky-high valuations but limited revenue models. He suggested advertising would become a primary revenue path for many AI companies. If that prediction holds, it creates an interesting dynamic: the same AI models that help advertisers build campaigns could also be selling ad inventory within their own chat interfaces.

Three developments will shape how this story evolves.

First, the beta results. When The Trade Desk eventually shares performance data from Claude-created campaigns versus human-created ones, the comparison will either validate the approach or expose its limitations. Expect that data sometime in the second half of 2026.

Second, Anthropic's positioning. Claude is currently being used as a campaign creation tool, but Anthropic hasn't announced a formal advertising partnership with The Trade Desk. Whether this evolves into a deep product integration or remains an experimental layer will determine its staying power.

Third, the OpenAI connection. Green declined to comment on reported talks between The Trade Desk and OpenAI. If The Trade Desk is simultaneously exploring both Claude and ChatGPT as campaign creation interfaces, the company may be building a model-agnostic approach that lets advertisers choose their preferred AI partner. That would be a smart strategic move, but it also suggests the current Claude integration may not be exclusive.

The future Green is describing isn't ten years away. It's happening now, in a closed beta that a small number of advertisers are already using. When it opens up, the way you build campaigns will change. Not might. Will. The only question is how fast.