AIX Inc. Representative Director Kiyoshi Inoue to Speak at "AI Engineering Summit Tokyo 2026"

AIX Inc. (Headquarters: Shibuya-ku, Tokyo; Representative Director: Kiyoshi Inoue), a developer and provider of AI solutions to enhance corporate productivity, spoke at the "AI Engineering Summit Tokyo 2026" conference hosted by Findy Tools on June 9, 2026.

Representative Director Kiyoshi Inoue delivered a presentation titled "The Story of Building an AI Agent for Autonomous Operation of a 200 Million Yen Monthly Advertising Budget Over One Year." He shared a case study of implementing and operating an AI agent in the advertising operations field as a year-long, hands-on FDE (Forward Deploy Engineer). He also introduced "LoreOS," an AI for executives, which is a productized version of the insights gained from this case.

Background of the Presentation

While the utilization of generative AI is a critical management issue, many companies face challenges such as "stalling at the PoC (Proof of Concept) stage," "lack of adoption in the field," and "inability to quantify results." AIX engages with clients at the management layer, developing and implementing systems in the field to drive revenue, accompanying clients through their DX journey to achieve results as a "year-long, hands-on FDE."

In this lecture, he candidly revealed the design process and incidents that occurred during the transformation of an advertising operation, which previously involved about 20 people manually managing a monthly budget of 200 million yen and experiencing several million yen losses annually due to accidents with CPA (Cost Per Acquisition) ten times the expected rate, into an autonomous operation driven by an AI agent within one year.

Lecture Overview

The lecture explained the one-year advertising operations reform undertaken by AIX with a single engineer, phase by phase. Key points include:

1. Starting with "Definition" (Deliberately Avoiding AI Initially)

KPIs such as "high-quality budget spending" and "poor budget spending (= blowout)" were verbalized, unifying the operational judgment criteria that had been individualized.

A proprietary management screen was internally developed to forcibly stop operations when CPA exceeded a certain threshold, visualizing all media and campaigns on a single screen. Before AI implementation, monthly blowouts (poor budget spending) of 20 million yen were reduced.

2. Multi-Layered Defense Design Based on Real-World Incidents

Learning from an incident where "a campaign that should have been stopped continued to run," a multi-layered stopping design was adopted, including per-person stop logic, global forced stop, automatic Slack alerts, and final human monitoring.

The design philosophy, based on the premise that human monitoring capabilities inevitably decline behind automation, was shared.

3. Expanding Definitions with AI (Phase 2)

With the principle that AI comes into play only after definitions are established, AI models were built to generate hypotheses and return results, establishing a system where humans make the final approval.

Systematization increased the number of campaigns managed per person by approximately 1.5 times, evolving operations beyond long-held field intuition.

As a result of these efforts, it was reported that monthly blowouts (CPA exceedances) were reduced by 20 million yen, staff were freed from 24-hour monitoring, and a total revenue increase of approximately 500 million yen was achieved over the year.

The presentation materials can be viewed here: https://speakerdeck.com/shiori01/findy-ai-summit-zi-liao-google-suraito

【Main Achievements Over One Year】

20 million yen/month

Reduction in blowouts (CPA exceedances)

24 hours

Liberation from constant human monitoring

Approx. 500 million yen

Total revenue increase over one year

About the Executive AI "LoreOS" Introduced in the Lecture

At this lecture, AIX introduced "LoreOS," an enterprise knowledge platform that productizes the on-site knowledge gained. LoreOS is a system where AI automatically aggregates and organizes daily communications and documents scattered across various internal tools such as Slack, Notion, and Gmail into "company memory."

Unlike traditional conversational AI, which are "passive systems that only react when a human asks a question," its key feature is that AI continuously and autonomously organizes internal information. It resolves the stress and time spent by executives on confirming "What was that discussion again?" and digging up memories, enabling prompt management decisions and operational efficiency by instantly retrieving necessary information.

"LoreOS" Service Site: https://loreos.jp/

Comment from Representative Director Kiyoshi Inoue

"Before AI, there is 'definition.' Human judgment is ambiguous and not always correct. Therefore, instead of introducing AI immediately, it is crucial to first identify the operational judgment criteria in the field without AI, and then have AI expand upon it—this sequence determines success. Future engineers will not be 'builders' but individuals who understand the field, determine key variables, systematize definitions, and pass them to AI. AIX will continue to deliver talent and systems that can provide end-to-end support from field understanding to results generation."

About "AI Engineering Summit Tokyo 2026"

"AI Engineering Summit Tokyo 2026" is a conference hosted by Findy Tools, themed around AI engineering. It serves as a platform for sharing practical knowledge in AI-driven development and operations, with speakers and participants from companies and engineers both domestically and internationally.

Event Official Website: https://ai-engineering-summit-tokyo.findy-tools.io/2026-summer

AIX Inc. Company Profile

AIX Inc. is an AI consulting company that supports the development and implementation of specialized AI agents optimized for corporate operations. It provides end-to-end support from business analysis and requirements design to AI agent development, implementation, and operational adoption, realizing increased corporate productivity and DX promotion. It offers "implementation-type AI DX" that delivers results in the field, not just at the conceptual stage.

AIX Inc. Location: 2-23-2 Jingumae, Shibuya-ku, Tokyo, Harajuku Dress Building Representative: Kiyoshi Inoue, Representative Director Business Activities: AI-DX Support Business URL: https://aixinc.co.jp/

Media Inquiries Regarding This Matter

AIX Inc. Public Relations Email: [email protected] Inquiry Form: https://aixinc.co.jp/contact/