Neng Nana
By Krystal Scanlon
May 18, 2026
The Trade Desk has long positioned itself as the architect of the modern programmatic landscape, building a reputation on the back of complex, granular control. For years, the company’s platform has been the professional playground for expert media traders—those who thrive on manipulating thousands of variables to squeeze marginal efficiency out of digital ad spend. However, in an era defined by the rapid ascent of generative and agentic AI, The Trade Desk is making a calculated pivot.
The company is currently rolling out "Koa Agents," a new suite of capabilities designed to automate significant portions of the programmatic workflow. By integrating with third-party large language models (LLMs) through an interoperability layer dubbed the "Open Agentic Kit," The Trade Desk is signaling a fundamental shift: from a platform that requires expert intervention to one that actively assists, suggests, and builds on behalf of the user.
The Mechanics of Koa Agents: A New Workflow Layer
At its core, Koa Agents represents a strategic evolution of The Trade Desk’s existing "Koa" AI, which has been embedded within its Kokai campaign management platform since 2023. While the original Koa focused on optimizing campaigns already in flight, the new agentic layer shifts the focus to the "pre-flight" phase: strategy, setup, and troubleshooting.
The system is designed to be model-agnostic. Through the Open Agentic Kit, agencies and brands can plug in their preferred AI models—including Anthropic’s Claude—to interface directly with The Trade Desk’s ecosystem.
A typical workflow, according to industry insiders familiar with the beta, functions as a bridge between human intent and platform execution. A user can upload a raw media plan or a set of business objectives to a model like Claude. The agent then reformats this information into a template compatible with The Trade Desk’s architecture, essentially "building" the campaign structure automatically. Beyond mere data entry, the agents act as an operational co-pilot, capable of identifying creative issues, troubleshooting deployment errors, and providing detailed reasoning behind their recommended optimizations.
"We introduced Koa Agents as our effort to introduce agentic capabilities across our entire platform, everything from identifying and building audience strategies all the way to campaign setup, deployment and troubleshooting," explains Jordan Rost, vp of product marketing at The Trade Desk.
Chronology: From Ambition to Execution
The path to this moment has been characterized by a mix of high-level ambition and measured delivery.
- 2023: The Trade Desk launches "Kokai," its revamped platform architecture, placing the original Koa AI at the center of its automated decision-making.
- Early 2026: Development intensifies around the Open Agentic Kit, with the company engaging in closed-door conversations with key agency partners to test the viability of third-party model integration.
- March 2026: During the Marketecture Live conference, CEO Jeff Green offers a candid, unscripted glimpse into the future. When asked if users could leverage models like Claude to build campaigns, Green confirmed: "If you’re a part of our closed beta, yes," before jokingly adding, "I wasn’t supposed to say that."
- May 2026: The Trade Desk formally begins broader outreach to agency executives to socialize the Koa Agents framework, marking the transition from experimental R&D to active product rollout.
The Logic of Complexity: Why Programmatic Needs AI
Jeff Green has been a vocal proponent of the idea that programmatic advertising is uniquely suited for agentic AI. Unlike static creative production, programmatic buying is a high-velocity, high-variable environment.
"When you take a $500,000 campaign and want to turn it into a million-dollar campaign, you could change 10,000 different things," Green remarked at the Marketecture Live conference. "One of them, or two of them, or some blend of them is better than another. If agents can frame these choices and help you see the trade-offs, this is a perfect task for AI."
The sheer volume of variables—frequency caps, bid factors, geo-targeting, creative formats, and channel mix—creates a "decision-making headache" that is increasingly difficult for humans to manage at scale. The Trade Desk’s argument is that by offloading these tasks to agents, humans can move from being "knob-turners" to "strategy architects."
Crucially, the current iteration of Koa Agents is designed with a "human-in-the-loop" philosophy. The system does not operate as a black box; it requires marketers to sign off on decisions, and it provides an audit trail of the logic behind its suggestions. This focus on explainability is a direct response to the industry-wide anxiety surrounding AI systems that optimize for results without revealing their methodologies.
Data Portability and the Competitive Edge
A significant pillar of the Koa Agents rollout is data interoperability. Unlike some "walled garden" approaches that trap data within a platform to train proprietary models, The Trade Desk is emphasizing the ability to pull performance data out of its system.
"We want our system to really respect and feed that desire for brands and agencies to get smarter in the process," Rost noted. By allowing advertisers to export campaign performance data to train their own internal models, The Trade Desk is positioning itself as an open partner rather than a proprietary silo. In an agentic future where first-party data is the ultimate competitive asset, this move is a strategic attempt to ensure that brands remain tethered to The Trade Desk’s infrastructure while gaining the agility to build their own AI layers.
The Industry Paradox: Disrupting the Middleman?
The move toward automation presents a fundamental conundrum for The Trade Desk. As Shiv Gupta, founder of the staff-training service U of Digital, points out, the company’s historical success was built on the exact opposite of simplicity.
"The conundrum for The Trade Desk, and perhaps why they’ve been late to the game in this regard, is that simplifying programmatic buying is antithetical to their business model and their DNA," says Gupta. "They built their business by offering every knob and lever possible to traders… This worked particularly well for agencies, who were then able to charge their customers for managing programmatic campaigns on The Trade Desk."
By reducing the barrier to entry, The Trade Desk is effectively questioning the necessity of the massive "expertise layer" that agencies have historically provided. However, the commercial logic is compelling. In March 2026, the company signed 45 joint business deals—a 55% increase year-over-year. These deals represent a trend of brands seeking direct relationships with the platform, bypassing traditional agency intermediaries.
Strategic Implications: Who Wins in the Agentic Era?
The implications of this shift are profound for the entire advertising ecosystem.
1. The Holdco vs. Independent Divide:
Robert Webster, founder of the AI marketing consultancy TAU, suggests that the market is bifurcating. Large holding companies will likely build their own proprietary AI operating systems, creating bespoke competitive advantages. However, for smaller brands and independent agencies, the cost of such development is prohibitive. "MCP [Media Control Plane] plus Claude gives them a planning layer that works across Google, Meta, TikTok and the rest without the build cost," Webster explains. "That’s how The Trade Desk shores up the part of the market the holdcos don’t own."
2. The End of "Manual" Programmatic:
The shift suggests that the era of manual campaign optimization is approaching its sunset. As Koa Agents become more sophisticated, the role of the "trader" will likely evolve into that of an "AI systems manager," where expertise is defined by how effectively one can prompt, constrain, and validate the agents rather than manually adjusting bids.
3. The Risk of Homogenization:
If every brand and agency uses the same underlying agentic logic, will campaign performance become homogenized? The Trade Desk’s emphasis on data portability and third-party model integration is likely a defensive play against this. By allowing users to feed their unique, proprietary first-party data into the agents, the platform hopes to prevent the "race to the middle" that often accompanies mass-market automation.
Conclusion: A Delicate Balancing Act
The Trade Desk is currently engaged in a high-stakes balancing act. It is attempting to accelerate the automation of its own platform—thereby reducing the complexity that once defined its value proposition—while simultaneously working to maintain its status as the "neutral" backbone of the open internet.
By democratizing access through Koa Agents, The Trade Desk is betting that the stickiness of direct brand relationships will ultimately outweigh the revenue previously generated by the complexity-driven "expertise economy." Whether this move leads to a more efficient, creative-focused advertising landscape or creates a new set of dependencies remains to be seen. What is clear, however, is that the era of the autonomous, AI-augmented media plan is no longer a theoretical future—it is a live, beta-tested reality.
Lina Irawan
Laily UPN
