Pricing -70%  

Dynamic Residential

Static Residential

Wallet Top-up

EN
Console Logout
Sign up
Exclusive for New Users
Pricing
Dynamic Residential Dynamic Residential IP (Lines) Unlimited Residential Traffic Long-term Static ISP Wallet Top-up
Products
Dynamic Residential Traffic Unlimited Residential Traffic Long-term Static ISP United States France Canada South Korea Italy Japan Singapore
Get Proxies
Username/Password Mode-Dynamic Residential Traffic Username/Password Mode-Long-term Static ISP Username/Password Mode-Unlimited Traffic Bandwidth Username/Password Mode-Unlimited Traffic Port API Mode-Dynamic Residential Traffic
Resources
Affiliate Program FAQ Tutorials Blog Partners Global Resources
Language
ZH-简体中文 EN-English
Dashboard
Resource Blog Kokai Zuma: AI Agents Transform Ad Optimization

Kokai Zuma: AI Agents Transform Ad Optimization

1024blogadmin

2026-08-28

Key Takeaways:

  • What AI advertising capabilities Kokai Zuma adds and how Koa Assistant supports real advertising tasks.
  • What The Trade Desk’s reported 32% average CPA improvement means and how advertisers should interpret the figure.
  • How the growing role of AI in ad execution could change the way overseas advertising teams work.
  • How cross-border sellers can combine business data with AI to make smarter advertising decisions.

Imagine you are responsible for advertising an outdoor power station in overseas markets. You open your advertising dashboard and notice that the conversion rate of a campaign has suddenly dropped. You export the data and give it to an AI tool for analysis. Within minutes, it may point to several possible causes: excessive ad frequency, declining creative performance, or changes in traffic quality.

But the report is only the beginning. You still need to log back into the advertising platform, check the relevant settings one by one, and adjust the campaign based on the analysis.

That is the workflow Kokai Zuma aims to change.

On August 27, The Trade Desk officially launched Kokai Zuma, a major upgrade to its Kokai advertising platform. For overseas advertising teams, this is more than another AI tool. It signals that some tasks traditionally handled by people are starting to be redistributed between humans and AI.

What Has Kokai Zuma Upgraded?

The Trade Desk is a programmatic advertising technology company whose platform is used for ad planning, buying, and measurement. Kokai is its AI-powered advertising platform, while Zuma represents a major upgrade to the system.

On the surface, Zuma adds capabilities such as an AI assistant, audience building, and automated adjustments. But the more important change lies underneath these features.

The upgrade strengthens the platform’s prediction engine, AI infrastructure, and real-time data processing capabilities. These technologies support functions such as forecasting advertising performance, analyzing campaigns, and adjusting ad delivery.

This means AI is no longer simply sitting inside a separate chat window. It is becoming more closely connected to the data and operational capabilities of the advertising platform.

In simple terms, AI used to work more like an analysis tool: advertisers gave it data, and it organized the information, analyzed performance, and provided recommendations. Zuma takes AI further into the advertising workflow, allowing it to participate in more stages of the process.

Koa Assistant Brings AI Into More Advertising Tasks

One of the more visible changes in this update is the expanded role of Koa Assistant. It can help advertisers with tasks such as campaign creation, audience building, performance analysis, and troubleshooting.

This is also where it is important to distinguish generative AI from agentic AI.

Traditional generative AI generally follows a simple process: a person asks a question, and AI analyzes the information before providing an answer. For example, an advertiser can give an AI tool campaign data and ask why the conversion rate has fallen. The AI may identify possible causes, but deciding what to change, finding the relevant settings, and applying the changes still requires human intervention.

Agentic AI focuses more on completing a broader task. A person provides a business objective, and the AI can interpret the task, access relevant advertising data, analyze the situation, and use available platform tools to handle parts of the workflow before continuing based on the results.

In advertising, the workflow therefore starts to look different.

Previously, advertisers would collect campaign data, ask AI to analyze it, review the recommendations, and then manually make the changes in the advertising platform.

The emerging model is closer to this: define the advertising objective, analyze the available data, identify problems, develop an approach, execute certain adjustments, and then review the resulting performance.

This does not mean AI can completely take over an advertising account. The scope and level of automation still depend on platform permissions and product availability, while important advertising decisions still require human oversight.

The real change is that the gap between an AI recommendation and an actual advertising action is becoming smaller.

What New Advertising Capabilities Does Kokai Zuma Bring?

Zuma’s update covers several areas, including audience building, campaign creation, campaign optimization, troubleshooting, and performance measurement. Its value is not simply that one feature is particularly new, but that these capabilities are becoming more closely connected across the advertising workflow.

AI-Assisted Audience Building

Audience selection has traditionally been one of the more time-consuming parts of programmatic advertising. Advertisers need to identify potential audiences based on product characteristics and historical data, then repeatedly combine, test, and refine different segments.

If the audience is too narrow, reach may be limited. If it is too broad, the campaign may attract a large amount of low-value traffic.

Koa Assistant can use the platform’s data resources to help advertisers discover relevant audiences and expand existing targeting strategies.

For example, imagine an overseas company promoting an outdoor power station in the US market. It may initially focus on audiences interested in camping and RVs. As campaign data accumulates, AI can help identify additional audience directions based on existing performance data.

The real time savings here do not come from a single click. They come from reducing the amount of manual audience screening, combination, and testing.

However, AI-generated audience suggestions still need to be evaluated against product pricing, market demand, and profit margins. An audience that is likely to convert is not necessarily an audience worth targeting over the long term.

AI-Assisted Ad Frequency and Campaign Optimization

Ad frequency is another small but important advertising metric. If frequency is too high, the same user may see an ad too many times, wasting budget and creating ad fatigue. If it is too low, users may not see the ad often enough to remember the brand.

In the past, advertisers had to monitor historical and real-time performance, decide whether adjustments were necessary, and then make those changes manually.

Koa can use campaign objectives and real-time performance data to adjust advertising delivery. This type of work involves large amounts of data and frequent changes, making it well suited to AI-assisted automation.

However, overseas sellers should remember that improving advertising metrics does not necessarily mean improving business results. The effectiveness of AI should not be judged by one or two advertising metrics alone.

Campaign Creation and Troubleshooting

Creating a campaign also involves many repetitive tasks. Koa Assistant can help advertisers create campaigns based on their objectives, reducing the amount of repetitive configuration required when launching a new campaign.

Troubleshooting addresses another common problem. When spending becomes abnormal, impressions decline, or conversion performance suddenly changes, advertisers may need to check budgets, audiences, campaign settings, and traffic performance across multiple dimensions.

If AI can quickly analyze the available data and narrow down the potential causes, advertisers do not have to manually inspect every part of the dashboard.

This also means AI is becoming involved earlier in the advertising process. Instead of analyzing only what happened after a campaign ended, AI can participate in campaign creation, analysis, and troubleshooting while advertising is underway.

Advertising Measurement and Reporting Upgrades

Zuma also upgrades advertising measurement and reporting capabilities, including conversion lift analysis, advertising reports, report-building tools, and campaign setting views.

These features may appear less noticeable, but they are important when AI becomes involved in campaign adjustments.

Once AI starts making or recommending changes, advertisers need to know what was changed, why it was changed, and whether the change actually produced better results.

Only when these changes can be tracked and reviewed can AI-assisted advertising avoid becoming a black box.

32% Average CPA Improvement: AI Optimization Enters the Testing Stage

Beyond the new features, The Trade Desk has also released a performance figure worth watching.

The Trade Desk said its platform analysis showed an average 32% improvement in CPA after Koa optimizations were enabled. The company’s official wording is “average 32% improvement in CPA performance.”

This figure should be interpreted carefully.

First, the data comes from The Trade Desk’s own platform analysis rather than an independent third-party study. It should therefore not be interpreted as meaning that every advertising account can achieve a 32% improvement in CPA.

Second, a lower CPA does not automatically mean higher profits.

For example, suppose an overseas company reduces the CPA for a product from $30 to $20. On the surface, advertising efficiency has improved significantly. But if logistics costs increase at the same time, return rates rise, or the resulting orders have low profit margins, the company’s actual profit may barely improve.

For overseas sellers, the most important point about the 32% figure is therefore not whether they can reproduce the same result.

The bigger signal is that AI is moving beyond analysis and recommendations and becoming more involved in real advertising execution, with results that can be measured.

How Should Overseas Sellers Respond to AI-Driven Advertising?

As more AI advertising tools become available, the key question is no longer simply whether to use AI. The more important questions are which tasks should be handed over to AI and how to prevent AI from executing the wrong strategy more efficiently.

The core principle is simple: AI can only make decisions based on the data and objectives it has access to.

If AI only receives advertising data, it will naturally focus on advertising metrics. To make AI decisions more closely aligned with actual business performance, sellers need to provide a broader business context.

The following practices can help overseas sellers build a more effective human-AI workflow.

Define the Advertising Objective Before Using AI

AI optimization needs a clear objective.

Is the campaign intended to generate leads, increase conversions, control acquisition costs, or test a new market? Different objectives require different optimization approaches.

Before using AI, define the primary goal, the key metrics, and what success looks like.

For example, when testing a new market, sellers may initially care more about traffic quality and conversion quality. Once the campaign becomes stable, cost control and return on ad spend may become more important.

The clearer the objective, the easier it is for AI to work toward the intended outcome.

Connect Advertising Data With Real Business Data

This is particularly important for overseas sellers.

Advertising dashboards typically provide information such as impressions, clicks, spending, and conversions. But these metrics cannot fully determine whether an order is actually profitable.

A more complete evaluation system should combine multiple types of data.

Advertising data helps evaluate campaign efficiency. Website and sales data help determine actual conversion quality. Product costs, logistics, duties, and returns help calculate real business profit. Market and competitor information provides additional context about the external environment.

For overseas companies operating in multiple markets, public websites, market information, and competitor research can also provide additional data beyond what is available inside an advertising platform.

Using an IP service such as 1024Proxy can help businesses collect publicly available market information from different locations, supporting product research, competitor analysis, and advertising strategy decisions.

Avoid Using the Same AI Strategy Across Every Market

For companies operating across multiple overseas markets, data quality, competition, and consumer behavior can vary significantly from one country to another.

Instead of putting every market under the same automated rules, evaluate each market separately.

Markets with sufficient historical data and stable conversion performance may be suitable for a higher level of automation. New markets with limited data should be monitored more carefully before expanding automation. In some markets, a good advertising CPA may still hide weak overall profitability, requiring the seller to reconsider the strategy using broader business data.

The goal is not to restrict AI, but to avoid excessive automation where data is insufficient or the underlying business assumptions have not yet been validated.

Establish a Regular Human Review Process

A safer workflow can be:

AI analysis → review the supporting data → human assessment → limited execution → monitor results → expand automation if validated

Major changes involving budgets, audience targeting, or new-market testing should continue to receive human oversight.

If a platform provides different levels of automation, sellers can start with lower-risk functions, evaluate the results, and gradually expand automation as confidence increases.

The goal is not to distrust AI. It is to make sure automation is built on real-world evidence.

Shift Human Time From Repetitive Tasks to Market Decisions

As AI takes over more data organization, anomaly detection, reporting, and routine campaign adjustments, advertising teams can spend more time on higher-value decisions.

These include determining which markets deserve additional budget, identifying which products should receive more advertising support, understanding why customers convert or leave, monitoring local competitors, and making sure advertising strategies align with inventory, logistics, and profitability.

These decisions shape the business itself, not just individual advertising metrics.

AI can process large amounts of data, but people still need to interpret what that data means for the business.

Turn AI Optimization Into a Continuous Cycle

AI optimization is not something you configure once and then leave alone.

Overseas markets change constantly. Competition, consumer preferences, seasonal demand, and product performance can all shift.

A more practical approach is to establish a continuous cycle:

Business data → AI analysis → campaign adjustment → actual business results → new data → another round of analysis

If the first round of changes does not produce the expected results, the seller should return to the original business objective and data rather than simply allowing the system to continue.

The value of AI does not come from making an advertising campaign perfect in one step. It comes from continuously improving decisions through repeated data feedback.

AI May Not Replace Advertising Professionals, but It Will Change Their Work

Kokai Zuma shows that AI is taking on an increasing share of data processing and advertising tasks.

In the past, advertising professionals spent significant amounts of time reviewing reports, investigating anomalies, adjusting parameters, and changing campaign settings. As AI takes over more repetitive work, people can devote more time to market analysis, budget allocation, product strategy, and performance reviews.

But this does not make human expertise less important.

As AI becomes involved in more stages of advertising, people need to become better at setting objectives, evaluating data, managing risk, and determining whether AI-generated decisions actually make business sense.

Advertising professionals will need to focus less on simply knowing how to operate an advertising dashboard and more on whether they can define the right objectives, provide reliable data, and judge whether AI decisions support real business needs.

AI can handle more data and execution. People remain responsible for objectives, strategy, oversight, and final judgment.

This is not simply about one replacing the other. It is about how advertising work is being redistributed between humans and AI.

Conclusion

The most important part of Kokai Zuma is not how many AI features The Trade Desk has added. It is the fact that agentic AI is moving deeper into the actual advertising workflow.

From audience building and campaign creation to campaign optimization, troubleshooting, and performance measurement, AI is becoming involved in more stages of advertising.

For overseas sellers, there is no need to hand every advertising task over to AI simply to keep up with the trend.

A more practical approach is to first establish clear advertising objectives, reliable business data, and a human review process, and then gradually delegate suitable repetitive tasks to AI.

The future of advertising may not simply be about who has the most advanced tools. It may increasingly depend on who can build an effective human-AI workflow faster.

AI may become increasingly fast at execution. But people still decide where it should go.

← Previous Post Next Post →

Residential IP Solutions

Reliable Residential IP Resources

Efficient Data Collection & Market Analysis

Get Started View Pricing

Online Service

Telegram

Sales Manager

@proxy1024

https://telegram.me/proxy1024

Sales Manager

@proxy1024_com

https://telegram.me/proxy1024_com

Private transfers are not accepted. Please make the payment through the official website!