The next phase of programmatic advertising may be less about giving media buyers more controls and more about deciding which controls they no longer need to touch. The Trade Desk is moving in that direction with Kokai Zuma, a major update to its Kokai advertising platform that combines agentic AI, campaign forecasting, audience discovery and simplified measurement.

The release, announced August 27, positions Zuma as more than a user-interface refresh. The Trade Desk says much of the investment has gone into the AI infrastructure beneath Kokai, where forecasting models predict inventory availability, model campaign outcomes and support Koa’s increasingly autonomous campaign-management capabilities.

For media buyers, programmatic advertising has become a problem of scale as much as a problem of reach. Campaigns now span display, video, audio, connected TV and other digital environments, while advertisers are expected to reconcile increasingly fragmented audience signals and prove that spending is producing measurable business outcomes.

The Trade Desk is betting that artificial intelligence can absorb more of that operational complexity.

Its new Kokai Zuma release brings agentic AI into the company’s demand-side platform (DSP), alongside a redesigned interface and a simplified approach to measurement and reporting. The goal is to let traders spend less time navigating campaign settings and reports and more time making strategic decisions about media performance.

That shift reflects a broader change underway across advertising technology. AI is moving from a tool that recommends optimizations to infrastructure that can increasingly execute them.

Gartner expects more than 70% of global advertising spend to flow through self-service advertising platforms in which AI materially influences media buying, cost and outcomes by 2028. The research firm also says 60% of marketing organizations with at least $1 billion in revenue will have piloted or put complex AI-agent workflows into production by the fourth quarter of 2026.

For DSPs, that creates a competitive race around who can turn AI into useful campaign infrastructure rather than simply adding a chatbot to an existing interface.

Koa becomes the operating layer

Kokai has used AI-powered forecasting since its introduction in 2023, but Zuma expands the role of Koa, The Trade Desk’s AI system.

According to the company, Koa analyzes more than 20 million ad impressions per second. Zuma introduces a conversational Koa Assistant, backed by specialized Koa Agents designed to help with campaign creation, audience development, troubleshooting and performance analysis.

The platform can also identify additional audiences, optimize campaign frequency and automate parts of campaign management.

That is significant because campaign optimization traditionally requires traders to monitor performance, identify a problem and then manually change targeting, bidding or frequency parameters. Agentic systems aim to compress that loop: identify the opportunity, determine the appropriate action and execute it within predefined campaign objectives.

The distinction matters for enterprise advertisers. Generative AI can make interfaces easier to interact with; agentic AI can potentially change how work gets done.

The Trade Desk is also tying those capabilities to its Audience Unlimited offering, which gives advertisers access to third-party data from more than 70% of global data providers available through its Data Marketplace.

Forecasting is the bigger story

The redesigned interface is arguably the most visible part of Zuma, but the more consequential change may sit underneath it.

The Trade Desk says it has upgraded Kokai’s forecasting engine and AI infrastructure to predict available inventory and model campaign outcomes in real time. Initial results cited by the company show an average 32% improvement in CPA performance when Koa Optimizations were enabled.

That figure should be viewed as an initial platform result rather than a universal benchmark. The release does not establish that every advertiser will see a 32% improvement, and performance will inevitably vary by campaign objective, inventory mix, audience and measurement methodology.

Still, the number illustrates where the strategic value of AI in a DSP increasingly lies: not necessarily in generating text or images, but in making billions of small decisions about inventory, audiences, bids and delivery.

Measurement gets a redesign, too

Zuma also attempts to address one of programmatic advertising’s persistent weaknesses: translating media activity into understandable business outcomes.

The updated platform includes enhancements to Conversion Lift, a redesigned reporting homepage and a more flexible Report Builder. Advertisers can customize reporting dimensions and access saved or generated reports more directly.

That puts measurement closer to the center of the buying workflow rather than treating reporting as a separate operational task.

For enterprise media teams, this could be particularly relevant as advertising budgets increasingly cross CTV, digital video, audio and display. A DSP that can connect optimization decisions with outcome measurement has an opportunity to reduce some of the fragmentation between media buying and analytics.

The competitive DSP landscape

Zuma arrives as The Trade Desk competes with platforms including Google Display & Video 360, Amazon DSP, Yahoo DSP and other programmatic buying platforms.

The competitive distinctions are important. Google’s DSP benefits from access to the broader Google advertising ecosystem, while Amazon DSP can leverage Amazon’s commerce and first-party shopping signals. The Trade Desk’s positioning is different: it emphasizes independent access to the open internet and cross-channel buying rather than ownership of a major consumer media ecosystem.

That makes AI efficiency particularly important to The Trade Desk’s proposition.

Industry analysis cited in a 2026 SEC filing estimates that Google DV360, Amazon DSP and The Trade Desk together represented roughly 86% of DSP market share based on 2025 programmatic spending patterns.

As these platforms become increasingly automated, the competitive question will move beyond inventory access and bidding algorithms. Advertisers will increasingly evaluate how much strategic work a DSP can automate, how transparent those decisions remain and whether AI optimization can be independently measured.

What Zuma means for advertisers

For enterprise advertising teams, Zuma is ultimately a workflow bet.

A trader could use Koa Assistant to investigate campaign performance, identify a new audience, adjust frequency or explore reporting without navigating multiple layers of platform settings. Bulk-editing improvements also let users preview the potential effect of changes before applying them.

That could reduce operational overhead, but it also raises an important question for buyers: how much control should advertisers delegate to an AI system?

The answer will depend on transparency, approval mechanisms, auditability and the ability to understand why an optimization was made. As AI takes on more campaign decisions, those governance features become just as important as performance gains.

Zuma is rolling out globally to Kokai clients, with some capabilities generally available and others in open or closed beta.

The broader signal is clear. The DSP is evolving from a software interface where traders manually manage programmatic buying into an AI-assisted operating layer that can increasingly predict, recommend and execute campaign decisions. The Trade Desk’s challenge now is proving that this automation consistently delivers measurable gains without turning the black box of programmatic advertising into an even larger one.

Market Landscape

The DSP market is entering a new phase in which AI optimization, measurement and automation are becoming core competitive differentiators.

The Trade Desk remains one of the major independent DSPs, competing with Google’s DV360 and Amazon DSP while taking a different approach from platforms that own large consumer ecosystems. The company’s advantage is its focus on open-internet buying across formats including CTV, video, display and audio, while competitors can offer proprietary data or inventory advantages.

Zuma reflects an industry-wide movement toward AI-influenced media buying. Gartner’s forecast that more than 70% of global ad spend could flow through AI-influenced self-service advertising platforms by 2028 suggests that AI will increasingly affect not only campaign operations but also pricing, targeting and outcomes.

For advertisers, the next competitive battleground will therefore involve AI transparency, measurement quality, automation controls and outcome-based optimization, rather than simply the number of channels a DSP supports.

Top Insights

• The Trade Desk’s Zuma adds agentic AI to Kokai, allowing Koa Agents to automate campaign creation, audience discovery, troubleshooting and optimization for media buyers.
• The platform’s upgraded forecasting infrastructure produced a reported 32% average CPA improvement, highlighting AI’s growing role in programmatic campaign decision-making.
• Zuma simplifies measurement through Conversion Lift enhancements, redesigned reporting and flexible Report Builder capabilities aimed at enterprise advertising and analytics teams.
• The release intensifies competition among The Trade Desk, Google DV360 and Amazon DSP as AI becomes increasingly central to programmatic buying and campaign performance.
• Enterprise advertisers will need to evaluate AI-powered DSPs not only on performance but also on transparency, governance, measurement and control over automated decisions.

Get in touch with our Adtech experts