TL;DR: The Trade Desk is running a closed beta where advertisers create programmatic campaigns by talking to Anthropic’s Claude. CEO Jeff Green confirmed it at Marketecture Live NYC on March 11, calling programmatic advertising “the most conducive industry in the world” for AI. No public launch date yet — but this signals the end of manually built campaigns.

How the Claude Beta Actually Works

The concept is deceptively simple: instead of clicking through The Trade Desk’s Kokai platform to set up targeting, budgets, and bidding strategies, beta advertisers describe what they want in plain English. Claude interprets the instructions and configures the entire campaign inside Kokai.

Think about what that replaces. A typical $500,000 programmatic media buy involves what Green described as “10,000 different things” that could affect performance — audience segments, inventory sources, bid strategies, frequency caps, brand safety settings, dayparting, device targeting, and dozens more. Today, experienced traders spend hours configuring these variables. In the beta, Claude handles it in a conversation.

The technical backbone is OpenTTD, a unified developer portal launched March 4 that consolidates UID2, EUID, OpenPass, OpenAds, and OpenPath under one access point. This infrastructure was clearly built with AI agents in mind — give them one clean API layer instead of five different systems.

The Real Debate: Speed vs. Expertise

Not everyone is excited. When The Trade Desk force-migrated advertisers to Kokai in August 2025, the pushback wasn’t that the AI was wrong — it was that experienced traders couldn’t verify or override its decisions at scale. Their years of expertise were being displaced by a system that made implicit choices they couldn’t see.

The Claude beta amplifies this tension. When a language model configures your campaign, which default settings does it apply? How much of the trader’s institutional knowledge survives the translation from “run a brand awareness campaign targeting millennial homeowners” to an actual Kokai configuration? And when the campaign underperforms, who diagnoses the problem — the human who can’t see the settings, or the AI that chose them?

Green’s counter-argument is compelling: programmatic’s complexity is exactly why AI should handle it. Humans can’t process 10,000 variables simultaneously. AI can. The question isn’t whether AI will be better at configuration — it’s whether advertisers will trust it enough to let go.

The Bottom Line

The closed beta has no public launch date, but Green hinted at a full agentic AI framework for partners coming Later in 2026. If you’re running programmatic campaigns on The Trade Desk, start documenting your campaign setup processes now — your briefs, your targeting logic, your optimization rules. When this rolls out broadly, the advertisers who can clearly articulate their strategy to an AI will outperform those who relied on muscle memory and manual tweaking.

Key Insight: This isn’t just about The Trade Desk. Amazon is opening its ad stack to AI agents too. The entire programmatic industry is moving toward AI-first campaign creation. The traders who adapt become AI strategists. The ones who resist become obsolete.

For a deeper look at the tools shaping this space, see our Best AI ad copy generators 2026 guide.