Manufacturing AI Logistics Conference 2026 in Hiroshima | Detailed Keynote Report

What is the greatest change AI will bring to logistics? Perhaps it is not simply greater operational efficiency or optimization, but a future in which AI makes autonomous decisions, coordinates the differing objectives and interests of companies, and moves logistics forward across corporate boundaries.

On October 14, 2026, Masaya Higashi, Representative Director and CEO of ONZALINX Inc., delivered a keynote address at the “Manufacturing AI Logistics Conference 2026 in Hiroshima,” held at the Hiroshima International Conference Center.

In his presentation, Higashi outlined the possibilities created by autonomous AI-driven logistics and the structural challenges that prevent companies from collaborating. Recognizing that overall optimization alone is insufficient to make collaborative logistics work in practice, he proposed a new logistics framework in which AI negotiates the different objectives of individual companies and verifies fairness and the rationale for participation before execution.

He also presented a demonstration of collaborative-logistics negotiations among AI agents representing three companies, illustrating a vision for intercompany AI logistics that combines the verification of decisions through AI assurance with human approval.

At the end of the keynote, Higashi announced “HIROSHIMA FLOW,” a vision for making Hiroshima a hub for demonstrating the Physical Internet in the age of AI. He set out a new direction for connecting global discussions on AI governance with local, real-world implementation.

Technological Choices Made 100 Years Ago Still Shape Society Today

At the beginning of his presentation, Higashi used Japan's electrical power frequencies to explain the relationship between the adoption and standardization of technology.

In Japan, eastern Japan uses 50 Hz electricity, while western Japan uses 60 Hz. This difference dates back to the Meiji era, when the eastern and western regions introduced different types of generators, and it remains more than 100 years later.

Technological choices that were reasonable for each region at the time went on to have long-term consequences for society as a whole.

Higashi pointed out that a similar challenge could arise with AI. Even if individual companies introduce their own AI systems and optimize their respective operations, that does not necessarily mean society as a whole will become interconnected.

The important thing is not to consider how to connect AI systems only after they have become widespread, but to design mechanisms that enable AI systems to cooperate across corporate and organizational boundaries from the outset.

“How we incorporate AI into logistics today may determine what logistics looks like 100 years from now.”

This question served as the starting point for the entire keynote.

---

The Biggest Change Brought by AI Is Not “Optimization,” but “Autonomy”

When people think about using AI in logistics, applications such as delivery-route optimization, demand forecasting, and warehouse automation often come to mind.

Higashi, however, explained that the greatest change brought by AI lies not in making existing operations more efficient, but in “autonomy.”

Conventional logistics systems were designed to accurately carry out decisions made by humans. By contrast, as AI agents become more widespread, a new possibility is emerging: AI itself can assess situations, negotiate with other AI systems, and proceed to execution when the necessary conditions are met.

In other words, logistics is beginning to shift from a stage in which “humans use AI to optimize operations” to one in which “AI systems connect with one another and move logistics across corporate boundaries.”

At the same time, the more autonomous AI becomes, the greater the possibility that it will make decisions or take actions that humans did not anticipate in advance. It is therefore essential to design what should be to AI and under what conditions execution should be permitted.

Higashi also referred to the spread of standard technologies that connect AI agents to external systems and data, as well as developments such as the Logistics Management Officer Collaboration Promotion Council (J-CLOP). He described how logistics is moving beyond efficiency improvements within individual companies toward collaboration among companies.

---

Connecting Data Does Not Mean Connecting Companies' Objectives

The Physical Internet, which the Japanese government aims to realize by 2040, is a vision for sharing logistics resources across corporate boundaries to improve the efficiency of logistics throughout society.

But if data and systems become technically interoperable, will collaborative logistics between companies naturally follow?

Higashi's answer to this question was clearly no.

The issue is not only that data formats differ between companies. Their business objectives and constraints also differ fundamentally.

One company may prioritize reducing logistics costs, while another places greater importance on meeting delivery deadlines or maintaining quality. For logistics providers, driver workloads and vehicle utilization rates are also important decision-making factors.

When companies have different objectives, they will evaluate the same logistics plan differently.

Higashi pointed out that enabling AI systems to connect technically and enabling companies to cooperate economically and strategically are two separate challenges.

“Technologies for connecting AI systems are being standardized around the world. But how do we connect the decisions of different companies?”

This question lies at the heart of ONZALINX's current research and development efforts.

Optimizing the Whole Does Not Necessarily Benefit Everyone

One of the biggest obstacles to collaborative logistics is how benefits are distributed among participating companies.

Higashi introduced a comment made by an actual truck driver during a social media discussion about collaborative logistics: “I can't shake the suspicion that one party will end up taking all the benefits.”

Even if collaborative logistics reduces overall logistics costs for society, that does not mean every company will benefit equally.

For example, delivery efficiency might improve while a particular logistics provider is left bearing increased waiting times or additional work.

In the keynote, Higashi identified obstacles to collaborative logistics, including the distribution of benefits, delivery times, physical conditions such as vehicles and cargo configurations, company-specific rules, and the allocation of responsibility for accidents and cargo handling.

He also referred to university research on the Physical Internet, explaining that even when sharing logistics resources improves overall efficiency, the benefits to individual participating companies do not necessarily improve.

This challenge cannot be sustainably resolved simply by asking companies to understand one another better or change their attitudes.

Each company has its own interests to protect, and prioritizing those interests is itself a natural business decision.

Higashi argued that, rather than relying on goodwill or self-sacrifice, collaborative logistics requires institutions and technologies designed to give each company a reason to continue participating.

---

Logistics AX Means Transforming How Companies Make and Evaluate Decisions

As a way forward, Higashi presented the concept of AX (AI Transformation) in logistics.

Logistics AX is not simply about introducing AI into business operations. It means using AI to transform the very mechanisms through which companies make and evaluate decisions.

In conventional logistics optimization, the primary goal has been to minimize predefined indicators, such as transportation costs or delivery distances.

By contrast, in intercompany AI logistics, multiple companies bring their different objectives and constraints to the table, and their respective AI agents negotiate with one another.

What is needed is not a single optimal solution, but an agreement that participating companies can accept and that enables them to continue cooperating.

Higashi proposed developing the Physical Internet beyond a network for sharing logistics resources into a trusted foundation that makes co-creation among companies possible.

To achieve this, mechanisms are needed to understand and translate each company's objectives, negotiate conditions, verify agreed decisions, and then execute them.

---

Three Companies' AI Agents Negotiate: A New Experiment in Intercompany Logistics

In the middle of the keynote, Higashi introduced a demonstration in which AI agents representing three companies negotiated collaborative logistics arrangements.

The distinctive feature of this demo was that AI did not merely calculate the most efficient delivery route. Instead, each company had different objectives, and its AI agent negotiated while protecting the company's own conditions.

Each company's AI agent presented logistics conditions, assessed proposals from the other agents, and explored whether collaborative logistics could be established.

The demonstration also assumed a change in demand caused by a disaster. Rather than maintaining the initial logistics plan, it showed how AI could detect the change and renegotiate based on new conditions.

The important point is not to automatically execute a plan once it has been created, but to continue updating decisions between companies as circumstances change.

The demo illustrated that even a proposal that improves overall logistics efficiency may concentrate additional burdens on a particular participating company.

Higashi asked the audience whether they would be able to continue participating in such a collaborative logistics arrangement if they were the manager of a logistics company facing an increased burden.

This question demonstrated concretely why overall optimization by AI alone is not enough to sustain cooperation among companies.

---

Do Not Execute AI Decisions Unconditionally: The Importance of VERIFY

As AI begins to negotiate autonomously across companies, how to verify the resulting decisions and determine whether they may be executed becomes increasingly important.

The demonstration introduced a mechanism for performing “VERIFY” (verification) on logistics proposals generated by AI.

Verification covers more than the efficiency of logistics as a whole. It also includes conditions intended to prevent participating companies from being disadvantaged, fairness, and authorization to execute.

Through AI assurance, the system checks whether predefined conditions have been satisfied and allows only proposals that pass verification to proceed to the next stage.

Human approval is then incorporated before final execution.

This differs from a system that adopts AI-generated answers unconditionally.

By separating the function that generates decisions from the function that verifies whether those decisions satisfy established conditions, the approach seeks to achieve both autonomy and safety.

Higashi emphasized the importance of making it possible for third parties to verify afterward who authorized the execution of an AI-involved decision, under what conditions, and on what grounds.

AI governance is not simply about improving AI performance.

It is about making the conditions, grounds, and chain of responsibility behind AI decisions verifiable, and creating a state in which AI can be used with confidence in actual business operations.

---

Six Principles for Intercompany AI Logistics

Higashi presented six principles as a new way of thinking about how to realize intercompany logistics in the AI era.

Different Purpose: Assume that each company has different objectives.

Minimum Disclosure: Do not require more information to be disclosed than necessary.

Participation Rationality: Ensure that each company has a rational reason to participate.

Agreed Fairness: Satisfy mutually agreed rules of fairness.

Verifiable Decision: Make the conditions and grounds for decisions verifiable.

Verified Commit: Execute based on agreements that have undergone verification.

These are not merely appeals to goodwill intended to encourage cooperation between companies. They are design principles needed when AI handles different companies' objectives and constraints.

Not every company needs to share the same objectives. Nor is it necessary to share all data, including trade secrets.

If each company can protect its own interests, has a rational reason to participate, and can confirm that agreed conditions are being followed, the possibilities for intercompany collaboration will expand.

The keynote emphasized the need to evaluate the success of collaborative logistics not only by overall cost reduction, but also by efficiency, fairness, and the ability to sustain participation.

---

What the World Needs Is Not Just Smarter AI

In the latter part of the keynote, Higashi also addressed the fact that autonomous AI-driven supply chain planning has not necessarily progressed easily.

Referring to research and forecasts from Gartner, he noted that although more companies are making substantial investments, significant challenges remain in automating complex planning decisions.

Unlike routine decisions such as replenishment and ordering, operations such as delivery planning and inventory policies involve numerous constraints and business judgments.

Rather than making AI adoption an end in itself, companies must therefore design which decisions to delegate to AI and where to retain human involvement and approval.

Higashi argued that what the world lacks is not only AI capability, but also mechanisms for determining “how far AI should be allowed to go.”

This idea is closely linked to the verification of decisions through AI assurance, agreement formation among companies, and the management of execution conditions.

---

Hiroshima AI Assurance Council: Putting Global Discussions into Practice in Hiroshima

As AI increasingly makes decisions across corporate boundaries and moves real-world logistics, mechanisms for ensuring safety and reliability become indispensable.

Against this backdrop, the Hiroshima AI Assurance Council was established in Hiroshima.

Higashi explained that the purpose of the council is not to restrict the use of AI, but to make more ambitious real-world implementation possible by creating a foundation on which AI can be used with confidence.

The council also participates in the Partners' Community of the Hiroshima AI Process.

The Hiroshima AI Process is a framework that advances international discussions on AI safety and trustworthiness. A diverse range of organizations, including OpenAI, Google, and the World Bank, participate in the Partners' Community.

Higashi emphasized the importance of connecting these international discussions on AI governance with concrete technological implementation in the manufacturing and logistics industries.

“Discuss it globally, implement it in Hiroshima.”

These words express the intention not merely to advocate international principles and guidelines, but to turn them into mechanisms that work in actual business transactions and logistics operations.

---

HIROSHIMA FLOW: An Intercompany AI Logistics Model Starting in Hiroshima

At the end of the keynote, Higashi presented “HIROSHIMA FLOW,” a vision for advancing the real-world implementation of intercompany AI logistics from Hiroshima.

HIROSHIMA FLOW is an initiative in which multiple shippers and logistics providers, each with different objectives and constraints, use AI to develop collaborative logistics proposals.

Its goal is not simply to share logistics data or make joint deliveries more efficient.

It aims to demonstrate a mechanism that translates differing objectives between companies, allows AI to negotiate conditions, verifies each company's rationale for participation and the fairness of the arrangement, and enables logistics decisions to be executed based on agreement.

Technologies such as mathematical optimization, AI agents, and AI assurance form the foundation of this approach.

Higashi also referred to the practical knowledge and decision-making capabilities that Japanese companies have developed over many years.

These strengths include the judgment of experienced professionals, the coordination skills of frontline managers, the experience needed to detect anomalies, and the ability to negotiate with business partners. Turning these strengths into algorithms, data, rules, and AI agents could create a new source of competitiveness for Japanese companies in the AI era.

The goal is to develop the practical knowledge of Hiroshima's manufacturing and logistics industries into technologies that can be used across companies, and to expand the resulting achievements throughout Japan and ultimately around the world.

HIROSHIMA FLOW is positioned as the first step toward that goal.

---

## The Future of Logistics Will Be Determined Not Only by AI Performance, but by “Mechanisms That Connect Companies”

This keynote broadened the discussion of AI-driven transformation in logistics beyond the conventional perspectives of operational efficiency and automation, toward mechanisms for intercompany decision-making and coordination.

Even as technologies for connecting data and AI agents advance, that alone does not mean collaboration between companies will succeed.

Companies have different objectives and interests to protect. Since overall optimization does not necessarily benefit everyone, mechanisms that ensure a rational basis for participation and fairness among participating companies are essential.

By combining AI-driven negotiation, mathematical optimization, verification through AI assurance, and human approval, it may be possible to build a new logistics foundation that allows different companies to cooperate with confidence.

The initiatives of the Hiroshima AI Assurance Council and HIROSHIMA FLOW also represent an attempt to connect global discussions on AI governance with real logistics challenges faced by regional industries.

Just as technological choices made more than 100 years ago continue to affect society's infrastructure today, the design principles we choose for AI now may have a profound influence on the logistics and industrial structures of the future.

Companies do not have to share the same objectives. They do not need to share all their data. Nor does overall optimization have to come at the expense of a single company.

If mechanisms exist to understand differing objectives, negotiate conditions, and verify decisions, companies can cooperate more closely.

What logistics needs in the AI era is not simply technology for moving goods efficiently, but technology for connecting the objectives and decisions of companies.

HIROSHIMA FLOW, an intercompany AI logistics model beginning in Hiroshima, represents an effort to bring that possibility into the real world of industry.