Why We Are Building the “TCP/IP for the AI Era” with ONZALINX

Partnering with a Company That Restored Operational Autonomy to the Field to Build a Common Foundation for AI Decisions

Who Do We Build the Infrastructure for the AI Era With?

Saying “We are building the TCP/IP for the AI era” is easy.

However, turning that statement into a system that actually functions in real-world corporate environments requires more than theory and technology. It demands a partner who deeply understands the daily struggles of frontline operations, where decisions are made, and what is accepted as formal operational practice — a partner with a proven track record of enabling companies to run their operations autonomously.

For us at GhostDrift Mathematical Institute, Inc., that partner is ONZALINX Co., Ltd.

ONZALINX is a company that has consistently implemented the principle of “returning operational autonomy to companies and their frontlines” in the logistics technology domain. In this article, we outline why we are advancing toward the construction of a next-generation common foundation together with ONZALINX.

1 | The Choice Not to Lock Customers into Proprietary Systems

A defining feature of INTER-STOCK, the logistics system platform developed by ONZALINX, goes beyond being an advanced Warehouse Management System (WMS). It lies in their commitment to fully disclosing source code to client companies and supporting the transition toward internal system management.

In conventional package software, companies are often forced to rely on vendors for modifications and maintenance, making it difficult to adapt systems at the speed required by operational teams. Fully opening source code and supporting internal maintenance carries the inherent commercial risk of reducing future vendor development and maintenance contracts.

Even so, ONZALINX chose this path, believing that enabling client companies to understand, operate, and continuously improve their own systems is essential for long-term organizational health.

Neither a pure vendor-dependent model nor a lock-in strategy, ONZALINX chose a management philosophy that prioritizes the autonomy of client companies and logistics operations.

2 | ONZALINX Stayed Involved Beyond Delivery — Through Operational Adoption

Granting software freedom to a company is one thing; ensuring it actually functions effectively in day-to-day operations is quite another. Simply delivering software according to specification documents does not create a system capable of handling the messy realities of the warehouse floor.

Rather than stopping at disclosing source code, ONZALINX has consistently formed unified teams with clients, frontline workers, and development partners, walking alongside them until systems are deeply rooted in daily operations.

A clear example of this approach is demonstrated in the case study of YAHATA CO., LTD., a global manufacturer and distributor of industrial fasteners (Reference Video).

Throughout the project, the team worked through initial internal concerns, changes in the project leadership structure, and operational alignment between Automated Guided Vehicles (AGVs) and warehouse routines. A project lead stepped in to coordinate frontline requirements, while engineers from the development partner, Sigma International Co., Ltd., joined on-site to work through real-world edge cases — such as return processing, shipping labels, and physical identification tags.

As a result, AGVs now perform automated shelf realignment overnight to prepare for the next morning’s tasks, and daily cycle counts have helped achieve near-zero inventory discrepancies within the AGV area. Furthermore, seeing the clean and efficient warehouse in action increased client visits, transforming what was once considered a back-office operation into a showcase facility.

At the core of this transformation was the consistent philosophy upheld by ONZALINX: “The user is the main character, and the frontline nurtures its own system.”

3 | The Same Structural Problem Re-emerges in the Domain of “Decisions” in the AI Era

Historically, many companies relied heavily on external vendors for system development, encountering a wall where software could not keep pace with business changes. In the AI era, the same structural risk is re-emerging in the domain of automated decision-making.

With advancements in Large Language Models (LLMs) and AI agents, environments that offer highly sophisticated recommendations for order volumes, inventory placement, and transport routing are becoming readily available.

However, in operational domains that involve product quality, safety, and contractual liability, no matter how accurate AI models become, AI outputs alone cannot be adopted directly as formal corporate decisions.

• Which company-defined business conditions are satisfied?

• Is there sufficient evidence (logs, verification data) to support the decision?

• At what stage should execution stop when an exception occurs?

• Where does the company formally assume responsibility for the decision?

Unless companies retain their own criteria and verification mechanisms internally, they risk regaining operational autonomy only to hand over “decision autonomy” to AI model providers and AI vendors.

4 | Why Collaboration with ONZALINX Is a Logical Necessity

ONZALINX is not merely an integration destination for embedding GhostDrift’s AI technology into an existing logistics system.

They are experts in unraveling frontline business requirements, translating them into system architecture, and creating environments where client companies can foster their own operations.

Conversely, GhostDrift’s “Responsibility OS” structures business conditions, required evidence, accountable entities, and authorization states into verifiable records. It uses an independent Verifier to confirm compliance with these conditions before allowing decisions to be formally adopted into corporate operations.

The collaboration between the two companies extends the frontline-first philosophy ONZALINX established in “operational execution” into the domain of “AI decision-making and adoption.”

• INTER-STOCK (Logistics Infrastructure)

• Reclaiming Autonomy: Preventing operational execution from being entirely vendor-dependent

• Frontline Role: The frontline nurtures its own systems

• Target State: Returning operational autonomy to the enterprise

• Responsibility OS / HAAP (AI Infrastructure)

• Reclaiming Autonomy: Preventing AI decisions from being entirely model-dependent

• Frontline Role: The enterprise defines its own adoption criteria

• Target State: Returning decision adoption rights to the enterprise

By combining ONZALINX’s process design and field implementation capabilities with GhostDrift’s mathematical and mechanical verification technologies, we were able to translate this concept into a concrete reference implementation.

5 | Moving Beyond Philosophy to Joint Implementation

This partnership extends beyond theoretical alignment; it encompasses joint technical development and implementation.

The two companies filed a joint patent application for technology that links AI-generated proposals with enterprise conditions, evidence, approvals, and accountability records, and manages whether those proposals may be adopted as formal organizational decisions.

Furthermore, the teams completed Phase 1 of a joint Proof of Concept (PoC) modeling pharmaceutical cold chain handoffs (Press Release), publicly releasing the reference code and verification environment.

In this PoC, a reference implementation was built to inspect evidence — such as temperature records and receipt confirmations — against pre-defined rules. The system generates one of four statuses: RELEASE (Adoptable), HOLD (Insufficient Evidence), QA_REVIEW (Additional Review Required), or REJECT (Conditions Unmet). When a status other than RELEASE occurs, evaluation of subsequent segments is automatically halted. Replay verification confirmed that the decision label and digest match when rerun with the same input, rules, and implementation.

This implementation was made possible through the synthesis of field experience and mathematical verification theory.

6 | Building the “TCP/IP for the AI Era” with ONZALINX

Beyond standardizing operations within a single organization, connecting decision-making across shippers, carriers, consignees, and autonomous AI agents requires a common format.

ONZALINX and GhostDrift are currently developing an initiative called HAAP (Hiroshima AI Assurance Protocol) to address this domain.

HAAP is being developed as an assurance and interoperability layer designed to record and exchange requirements, responsibility information, evidence, verification results, adoption states, and transition authorizations in a verifiable, standardized format. It does not replace transport or protocol layers like TCP/IP, gRPC, or MCP, but complements them by validating conditions for formal adoption. While operating at a different technical layer than basic network transport like TCP/IP, we refer to it as the “TCP/IP for the AI Era” because it aims to serve as a universal foundation for connecting decision logic across enterprises and AI systems.

A protocol of this nature cannot be designed effectively in an isolated research laboratory. It can only become a trusted social protocol when developed alongside a partner like ONZALINX, who has years of experience working directly on the warehouse floor to return operational control to enterprises.

Conclusion | Why We Partner with ONZALINX

Building a new foundation for the AI era requires more than just new AI models. It requires a partner who understands the friction that occurs after technology is introduced into the field, where humans make decisions, and what constitutes formal operational adoption.

ONZALINX chose not to lock clients into proprietary systems, opting instead to empower enterprises and frontline workers. They have built a solid track record of applying this philosophy across logistics sites in Japan.

From logistics where the frontline nurtures its own operational systems, to logistics where enterprises responsibly adopt AI decisions and connect them across organizational boundaries:

We are building this next foundation together with ONZALINX, originating in Hiroshima and expanding to the world.

Primary Sources and References

1. YAHATA CO., LTD. INTER-STOCK Case Study Video (ONZALINX Official YouTube Channel)

2. Video URL: https://www.youtube.com/watch?v=xi82LkB_8hE

3. Referenced Content: Core system challenges, AGV selection rationale, INTER-STOCK flexibility, one-team frontline alignment, achieving near-zero inventory variance, multi-site expansion plans.

4. ONZALINX Co., Ltd. INTER-STOCK Product & In-House System Management Policy

5. Referenced Content: Product philosophy focusing on supporting future internal system management, preventing vendor lock-in, and allowing user enterprises to retain control over logistics IT.

6. Articles and Press Releases by Seiya Azuma, CEO of ONZALINX

7. Referenced Content: Hybrid internal development philosophy emphasizing in-house capabilities in an era of rapid change, user-centric IT investment, and implementing as OneTeam between user and IT vendor.

8. Strategic Partnership Announcement: ONZALINX Co., Ltd. × GhostDrift Mathematical Institute, Inc.

9. Referenced Content: Strategic division of roles between logistics process design/field implementation (ONZALINX) and AI decision verification/Responsibility OS (GhostDrift).

10. Joint Patent Application Announcement: “Transforming Logistics AI Proposals into Actionable Corporate Decisions”

11. Referenced Content: Technical patent separating AI proposal generation from formal corporate decision adoption, linking business conditions, approvals, and audit logs.

12. Press Release: Completion of Phase 1 Pharmaceutical Cold Chain PoC by GhostDrift Mathematical Institute, Inc. & ONZALINX Co., Ltd.

13. Press Release URL: https://prtimes.jp/main/html/rd/p/000000005.000182721.html

14. Referenced Content: Reference implementation for handoff verification, 4-stage automated status generation (RELEASE/HOLD/QA_REVIEW/REJECT), automated execution halting on unmet conditions, and deterministic replay verification.

15. METI, AI Guidelines for Business Ver. 1.2 / G7 Hiroshima AI Process International Guiding Principles

16. Policy URL: https://www.meti.go.jp/shingikai/mono_info_service/ai_shakai_jisso/20260331_report.html

17. Referenced Content: Directions established in international and domestic guidelines regarding human-centric AI, safety, transparency, accountability, traceability, and appropriate human involvement/oversight.