AI Analyzes Winning Ad Patterns and Adds Company-Specific Improvement Suggestions to 'Ad Operations AI Agent'

Only verified winning patterns are used as the basis for ad recommendations.

renue Inc. (Headquarters: Minato-ku, Tokyo; CEO: Yusuke Yamamoto; hereinafter "renue") has upgraded its 'Ad Operations AI Agent' to include only verified winning patterns as the foundation for daily ad improvement suggestions. Winning patterns refer to insights confirmed by humans based on actual campaign results. Unverified assumptions are not used as a basis for recommendations.

Overview of Ad Operations AI Agent Enhancement

renue has added a mechanism that uses only verified winning patterns as the basis for ad improvement suggestions. While the AI provides assessments, only humans can finalize and confirm these as verified.

The Ad Operations AI Agent was launched in February 2026. It integrates campaign planning, creative production, media submission, and post-delivery analysis on a single platform.

This update structurally incorporates 'evidence' into each suggestion. Daily generated recommendations now include the target campaign, supporting rationale, expected impact, success criteria, and the specific winning pattern used as the basis.

Suggestions require human approval before execution. Even if a user requests execution via the confirmation screen, ads are not changed immediately. The responsible operator performs the change and records both pre- and post-change values. Budget-related changes are designed not to be executed by AI or automated processes.

Additionally, the confirmation list has been streamlined to display only items requiring judgment. Observational data is displayed separately, allowing users to instantly see pending decisions and estimated response timelines.

Accumulated insights are stored exclusively within each advertiser's account. Customer ad accounts and delivery data are not used to train renue's AI models. Instead of training renue's AI, evidence accumulates directly within the customer's environment.

The more campaigns are run, the richer the decision-making basis becomes. Even if personnel change or external agencies are replaced, the evidence remains within the account.

Background and Objectives

Internet advertising operations involve separate management interfaces for each platform, with daily adjustments to bidding, budgets, and creatives. Since both cost-per-acquisition and delivery performance fluctuate, settings require continuous optimization.

With the rise of generative AI, more companies are integrating AI into ad operations. AI now handles creative ideation, report generation, and improvement suggestions, significantly reducing manual workload.

However, caution remains regarding blindly implementing AI-generated recommendations. Advertising budgets require accountability, and decision-makers must be able to explain budget allocations after the fact. As operations accelerate, tracking decision rationale becomes increasingly difficult.

Moreover, advertising know-how often remains siloed within individual staff members or external agencies. Insights on effective messaging or targeting are typically accumulated as personal experience. When staff rotate or agencies change, previous winning strategies are often lost. Even with continuous spending, the client company fails to build internal capabilities.

In February 2026, renue launched the Ad Operations AI Agent to address these challenges. Built on Anthropic's Claude Code, it consolidates tasks previously requiring multiple specialists and tools into a single operational platform. Supported platforms include Instagram, Facebook, Google Search, YouTube, X, and TikTok, with daily improvement suggestions generated based on Google Ads and Meta Ads performance data.

The initial version emphasized a cycle of analysis, hypothesis, creation, and verification. Post-launch operations revealed that the key to success lies not in accelerating the cycle, but in how effectively the results are preserved.

No matter how many unverified tactics are implemented, improvement direction remains unclear. Conversely, even verified insights become worthless if they don't inform future suggestions. Making multiple changes simultaneously makes it impossible to isolate what drove performance.

This enhancement aims to institutionalize this cycle. The system records three critical points: the rationale behind each suggestion, who approved it, and what changes were actually implemented.

renue aims to transform advertising operations from an 'external expense' into an 'internal asset,' enabling decision-makers to review evidence before making choices. Instead of having AI make decisions, the system supports human judgment with clear evidence.

Existing Challenges

Despite the growing adoption of AI in advertising, three major challenges remain in practice.

Lack of documented rationale for suggestions

AI-generated improvement suggestions often communicate only conclusions, without explaining why a particular approach is recommended or what evidence supports its effectiveness.

Without documented rationale, it's impossible to retrospectively evaluate suggestions. Explanations in board meetings or approvals rely on individual memory. This opacity can delay budget approvals.

As a result, decisions not to adopt AI suggestions accumulate. While tasks are automated, decision-making speed remains unchanged, leading to AI tools being used only for report generation.

Knowledge siloing and poor knowledge transfer

Insights on effective messaging or creative performance are often accumulated by individual staff. Even with identical budgets, results vary significantly by operator.

When staff rotate or agencies change, successors cannot access past test results and must repeat similar trials from scratch. Operational reproducibility remains low.

Unverified tactics may be passed down as unexplained best practices. Settings with no clear rationale persist indefinitely, and operational expertise never accumulates within the client organization despite ongoing fees.

Loss of visibility when delegating to AI

Delegating operations to AI makes change tracking difficult despite faster execution. Since advertising budgets are consumed daily, failing to identify which change triggered unexpected delivery delays response time.

Budget changes are particularly impactful—misconfiguring a spending limit can cause immediate losses. Yet, in many setups, the boundary between AI and human authority remains unclear.

Increased suggestion volume also creates burden. When judgment-required suggestions and observational data appear in the same list, review time increases. Suggestions targeting paused campaigns add further inefficiency. The broader the automation scope, the more critical control design becomes.

Key Features of This Solution

Only verified winning patterns are used as the basis for suggestions

Only insights verified through testing serve as the foundation for future suggestions.

After each advertising cycle, retrospectives are recorded, preserving hypotheses, implemented actions, results, learnings, and next steps. A/B test outcomes are saved as-is at the time of evaluation.

The system suggests whether a hypothesis is valid based on performance data, but only human operators can finalize verification. Automatic confirmation based solely on data aggregation is not allowed.

The next suggestion is