Internet

HWW

AI is already a supercar.
Time to build the highway.

HWW — Hyper-Web World Wide

A New, Subject-Centered Internet Architecture

What is HWW?

What the internet achieved, and what it left undone

The internet and the web dramatically extended humanity's ability to connect. We can send information across any distance, reach the world's knowledge, and use countless services.

But reaching information and achieving an intent are different things. To actually get something done, you must find the right counterparty, confirm its qualifications and trustworthiness, agree on terms, hand over the necessary information and authority, and receive the result. On today's internet this process breaks at the boundary of every site, app, and account. That is why a single change of address means visiting your bank, your card company, and your telecom provider one by one and entering the same information again and again.

It is because the upper structure of the internet developed around functions. People and organizations have gone into each service in search of functions, adapting themselves to the way each one works. Technologies that link services have steadily advanced. But there is still no universal upper structure that consistently connects intent to result across different Subjects and systems.

The gap agentic AI revealed

As AI began to find counterparties, negotiate, and transact on people's behalf, this gap could no longer be hidden. Agents must interpret and adapt to identification, authentication, and authorization schemes and invocation styles that differ from service to service. Relationships and meanings that are not brought into common correspondence are filled in each time by inference and translation. Like a supercar on a gravel road, much of the intelligence is spent compensating for the flaws of the road.

A bigger engine is not enough. The road itself must change.

The next generation of connection

The purpose of connection is value exchange between Subjects. Sharing news, forming relationships, and transacting all belong to it.

Humanity has removed the obstacles to this exchange one by one.

• Post and courier relays overcame distance.
• The telegraph shortened time.
• The telephone connected voices in real time.
• The internet let us reach the world's information and functions.

The remaining obstacle is finding the right counterparty, trusting it, and completing the exchange under legitimate authority. Today this work is broken at the edge of every service.

HWW is the next-generation network that removes this obstacle: a Universal Correspondence network. The agent network, in which AI agents act on behalf of Subjects, can unfold properly only on top of it.

HWW: a new internet architecture with the Subject as the reference

HWW moves the reference point of connection from functions to Subjects. It places Subjects — people, companies, institutions — at the persistent center, and lets the counterparties, information, and functions they need interlock according to their intent. HWW calls this Universal Correspondence.

A Subject is a person, company, institution, or community to which intent, authority, and results are attributed. AI agents and devices can be identified separately and act on a Subject's behalf, but they are not in themselves sources of authority.

To this end, HWW separates two things.

• Identification from functions. A Subject's existence and relationships are not tied to any particular site or app. Functions become means that can be connected and replaced when needed.
• Data from functions. A Subject's data and records are not appendages of a particular service. They carry on with the Subject at the center, even when services change.

Universal Correspondence is not integration. Banks keep their systems, hospitals keep theirs, and each nation keeps its own institutions. What HWW makes universal is not the system but the correspondence between systems.

The building blocks of HWW

Subject and Universal Key
People find one another through familiar External Identifiers such as telephone numbers or email addresses. HWW uses a dual identification structure that links these External Identifiers with cryptographic Internal Identifiers such as DIDs. The reference that keeps a Subject's identification and relationships continuous even when identifiers change or services move is what HWW calls the Universal Key. The Universal Key is not a single number, not a new national ID, and not a central registry.

Four core functions

• Identification: Confirms who it is and what qualifications and authority it holds.
• Intent & Matching: Finds the counterparties and functions that fit what is wanted.
• Negotiation: Aligns terms, scope, and authority before execution.
• Execution: Performs as agreed and returns the result to the Subject.

PDC (Personal Data Center) — the space where you control your own data and relationships
The Subject-side environment in which the Subject itself controls its data, relationships, and authority, and the services and agents it uses. Apps and devices may change; the relationships in the PDC carry on.

HCTP (Hyper-Context Transaction Protocol) — the common protocol of Universal Correspondence that links Subject, Intent, and Function
If HTTP is the common protocol for accessing information on the web, HCTP is the common protocol that brings into correspondence who is interacting with whom, with what intent, and under what trust and authority. It does not replace existing APIs and protocols; on top of them, it lets them understand one another in the same meaning.

Relationship to the existing internet

HWW does not replace the existing internet. Telecom networks, the web, apps, cloud, and AI all continue to be used. HWW is a logical upper architecture placed on top of them; it does not insert a new layer into the communications layer model. What changes is not the technology but the reference by which connection is aligned.

What HWW changes

• You can carry your identity, relationships, authority, and permitted data with you even when you change services and agents.
• The sign-up, verification, and data entry repeated at every service are greatly reduced.
• Service providers compete on quality, not on holding the gateway of a platform.
• Instead of filling the road's flaws with inference, AI spends its intelligence understanding and judging people's intent.

Frequently asked questions

Is HWW a new internet?
It does not replace the existing internet, but it is a new internet architecture in that it places a new, Subject-centered upper structure on top of it.

Must every service join a single platform?
No. Each system stays as it is; only the rules at the boundaries where they connect are aligned in common.

Is it a blockchain technology?
No. HWW is an upper architecture, not a specific technology. Distributed ledgers can be used as component technologies where needed.

Does it need AI to work?
No. HWW is designed to work without AI. AI is a capability that assists search and judgment.

Is HCTP a standard?
Not yet. It is Draft 1.0; technical validation and international standardization are being pursued.

HWW & Agentic AI

AI has begun to act

AI is changing from a tool that answers into an agent that acts. Agents find counterparties, negotiate with other agents, make reservations, and pay.

But being intelligent and being able to act legitimately are different things. Finding a counterparty does not mean it can be trusted, and being able to call a function does not mean having the authority to execute it. Exchanging messages does not mean both sides understand each other's intent, terms, and responsibilities in the same way.

In the end, anyone facing an agent asks five questions: Whom does it represent? What is it acting for? What authority has it been given? Is that authority still valid? To whom do the results and the responsibility return?

A supercar on a gravel road

Today's internet cannot answer these questions structurally. Authentication and authorization schemes are scattered across services, and the relationships and state between them are not expressed in common. So agents reconstruct them by inference every time: Is this site genuine? Does this person have authority? Is this reservation still valid?

This is the supercar on a gravel road. The more numerous and powerful agents become, the more the cost of compensating for the road's flaws piles up in computation, energy, and risk.

Reasoning ability cannot replace a structure that proves identity and authority.

The reference is the Subject, not the agent

An agent may be created for a single task or operated for a long time. But an agent's identification or capability does not in itself become the source of legitimate authority. HWW places the reference point at the Subject behind the agent.

The interface may be an agent, but the reference must be the Subject.

An agent's identity is confirmed through its relationship with the Subject it represents. Trust comes not from the agent's own claims but from the chain of Subject, qualifications, and delegation. Even after an agent disappears, the Subject, delegation, and execution history on which its actions were based can be verified and traced. Legal responsibility is determined under the applicable law and contracts.

Intelligence does not create authority

However intelligent an agent becomes, that alone does not expand what it may do. Authority arises only from explicit delegation with a defined purpose, scope, amount, and period. For example, an agent entrusted with purchases up to 500,000 won cannot exceed that limit even if it finds a better item; it must ask the Subject for approval again.

Within the delegated scope, it can work without asking for approval each time; it returns to the Subject only when it would go beyond that scope. It is designed so that, even if the AI stops, a person can take over the same work state and continue.

How agents work on HWW

Subject → delegation → Agent → HCTP → identification · trust → search · matching → negotiation → execution → results to the Subject

• Identification · trust: The counterparty checks not the agent's word, but the Subject behind the agent and the delegation relationship.
• Search · matching: It finds counterparties and functions that fit the Subject's intent, not entries in an agent list.
• Negotiation: It agrees only within the delegated limits and conditions.
• Execution · results: Execution beyond the scope of delegation is designed to be blocked, and results and records are linked to the Subject.

Relationship to existing agent technologies

Agent-to-agent communication technologies, technologies that connect agents with tools, decentralized identifiers and verifiable credentials, and existing APIs all continue in their roles. HWW does not replace them; it lets them be used within a single relationship: "for which Subject, under what authority, doing what."

The technology may stay the same. What changes is the reference point by which connection is aligned.

Watch the videoAI agent identity, trust, discovery, and accountabilityYouTube

HWW & AI Sovereignty

AI sovereignty includes sovereignty over connection

AI sovereignty does not end with having your own AI models or computing resources. Once AI takes on the real work of society and the economy, what matters is who decides whom to identify and trust, who may access data under what authority, and with whom an agent may connect and what it may do.

The moment people entrust shopping, finance, healthcare, and public administration to agents, the strongest agent platform can become the entrance to a nation's economy and public services. Once identities, records, and relationships accumulate inside it, people cannot switch platforms even if they want to.

Three things to protect in the realm of connection

• Freedom to choose AI: Identity, relationships, authority, and records carry over intact even when the AI provider changes.
• Data sovereignty: Data and the rights to use it are not tied to a particular service, but remain under the control of the Subject and the institution concerned.
• Connection sovereignty: Subjects and systems decide for themselves under what trust and authority, and on what terms and rules, they connect.

The key is reversing the structure. Instead of people and institutions entering platforms, people and institutions freely choose, use, and replace AI and services.

A model the telephone network has already shown

Each nation's telephone network keeps its own numbering plan, institutions, and operators, yet connects with the whole world as one through international standards. Calling from +82 to +1 does not require merging the two nations' telephone networks.

The national infrastructure model HWW proposes is the same. Each nation has its own correspondence infrastructure, and nations federate through common rules. Federation is not integration, and connection does not automatically transfer trust. What to recognize is decided by the receiving side.

National common correspondence infrastructure: social overhead capital for the age of intelligence

Like roads, power grids, and telecom networks, the minimal common part everyone relies on is kept as public infrastructure, and the services above it compete in the private sector.

HWW calls this common foundation the Public Correspondence Highway. Multiple operators run it jointly under public rules, and no operator owns the Highway as a whole.

This is not a government-run platform. The public infrastructure does not gather individuals' original data in one place, nor does it provide services directly. It handles only the minimal registration and verification information needed for connection and confirmation. It runs only the rules of connection and confirmation that everyone uses equally — the backbone of trust.

What the public side does — roads and traffic lights

• Subject continuity and identification links: based on existing identity systems and telecom number verification. No new national ID number is created.
• Confirmation of trust and qualifications: licenses, employment, business registration, and insurance eligibility continue to be attested by their existing issuers; only the way they are checked is made common.
• Common rules for neutral search and matching, and a public matching space
• Correspondence rules and registries, governance, and the gateway for federation between nations

What the private side does — the cars and shops on the road

• PDC services, AI agents, applied search and recommendation, and execution services such as payment, reservation, and delivery. Anyone who meets the criteria can join and compete.

Separating operation from oversight
Even if a few operators run the core for efficiency, the authority to oversee fairness and integrity is shared among multiple independent bodies. This system of checks keeps operators from changing the rules or distorting matching.

The path to deployment

1. Demonstration: Validate security and efficiency through regulatory sandboxes and local-government pilots, and establish national specifications.
2. Diffusion: Create initial demand by moving public document delivery and administrative authentication to HWW first, and support the participation of small businesses and merchants.
3. Federation: Agree with neighboring nations on the scope of mutual recognition of identity and qualifications, and widen connection through international standardization discussions.

Connecting with the world while keeping each nation's autonomy

The purpose of Universal Correspondence is not to make every nation use the same AI and the same database. It is for people, companies, and agents in different nations to connect when needed while keeping their own laws and institutions. Recognition of trust between nations, conflicts of law, and attribution of responsibility are challenges to be resolved through real institutions and agreements; HWW proposes a common foundation that can carry those agreements technically.

The national infrastructure model on this page is a deployment model proposed by HWW, not a system currently in operation.

HWW Timeline

Milestones — the road so far

Roadmap — next steps

• Development of the HCTP specification's supporting documents: message schema, security profile, conformance tests
• Minimal demonstration (PoC) and release of a reference implementation
• Release of the White Paper for Developers and an SDK
• International standardization contribution (ITU-T FG-TIDA)
• Public-sector pilot

The roadmap shows our direction; the schedule will be updated according to validation results.

HWW Publications

Documents setting out the principles, architecture, and specification of HWW. The Summary can be downloaded right away. For the full white paper and the HCTP specification, leave your email address and we will send you a download link.

HWW White Paper — Summary

Korean 22 pp. · English 23 pp. · PDFFree download

The philosophy, structure, operation, and implementation of HWW in broad strokes.

Korean PDF Download English PDF

HWW White Paper Draft 1.0 (full text)

Korean 770 pp. · English 675 pp. · PDFPreface and contents open · full text by email

The full white paper: foundational philosophy and the architecture of Universal Correspondence, architecture and correspondence mechanisms, operational architecture and interaction flows, and implementation, operation, applications, and the correspondence economy.

Read the preface

Toward Universal Correspondence

The advancement of human civilization can also be understood as a history of overcoming the constraints of connection while expanding its capacity. Humans have continually sought to connect across greater distances, at greater speed and scale, and at lower cost, while the objects and forms of connection have expanded from the movement of people and goods to energy and signals, and further to information and functions.

Roads (路) such as highways, railways, and sea and air routes reduced the constraints of time and distance on the movement of people and goods. Lines (線), represented by power grids and communication networks, largely freed the transmission of energy, signals, and information from the constraints of physical movement. As representative forms of such lines, the Internet and the World Wide Web provided a universal foundation through which information and digital resources around the world can be discovered, accessed, and used.

The ultimate purpose of connection, however, does not lie in movement, transmission, or access to information itself. It lies in enabling the needs and offerings of different subjects to meet, allowing necessary information to be found, understood, and evaluated, and mobilizing appropriate functions and actions so that diverse forms of value—including information, knowledge, services, rights, and relationships—can be exchanged and realized.

This white paper regards this as the next dimension of connection: the problem of 結 (gyeol)—the forming of the relationships through which value is exchanged and realized. 結 does not refer merely to the act of value exchange itself. It refers to a connection order through which a Subject's intent can meet an appropriate counterparty, the necessary information, functions, trust, authority, and execution can be brought into correspondence, actual relationships and actions can be established, and value exchange can be completed through the attribution of its outcomes back to the Subject.

Today, various common networks and broadly interoperable systems already exist within individual domains that participate in value exchange. What remains insufficiently established is a structure capable of spanning these domains and universally corresponding the entire process—from a Subject's intent and the discovery of an appropriate counterparty, through the verification of trust and authority, to the mobilization of information and functions, execution, and attribution of results—as a coherent connection order.

AI makes this structural gap increasingly visible. This is because AI is rapidly reducing human cognitive constraints in searching, selecting, analyzing, and evaluating information, while Agentic AI is going further, evolving toward the ability to act on behalf of humans by discovering counterparties, selecting and invoking functions, negotiating, and carrying out real-world actions and transactions.

Accordingly, the problem of connection now extends beyond whether information can be accessed. It includes whose intent is being represented, with whom an agent should connect, on what basis a counterparty may be trusted, what information and functions may be mobilized under what authority, and to whom the results of execution should be attributed.

Agentic AI did not create this gap. Rather, as intelligent systems begin to perform discovery, judgment, coordination, and execution that humans previously carried out directly, a gap that has long existed in the current connection structure is becoming an explicit architectural problem.

The evolution of human connection, the development of communications networks, and the emergence of Agentic AI have followed different paths, yet they ultimately point to the same structural question: How can the elements required for a Subject's intent to meet the appropriate counterparty, information, and functions—and ultimately be realized as action and value under the necessary trust and authority—be universally corresponded?

The existing Internet was also developed to serve humans. Its connection structure, however, has largely been organized around information and functions such as hosts, resources, services, and applications. Subjects have generally had to find the functions they need, enter the respective service domains, and adapt their identities and intents to the structures of those domains.

As a result, the process of value exchange is fragmented and frictional across multiple dimensions. Subject identification is divided among service-specific accounts; information and functions among separate websites, applications, and platforms; and trust and authority among different systems of authentication, credentials, and institutional rules. To complete a single intent, a Subject may have to move repeatedly across discontinuous domains—to identify itself, search, evaluate, establish trust, switch functions, and resubmit its information and authority.

This white paper does not regard separation or friction themselves as problems. Separation may be necessary for specialization, autonomy, diversity, competition, security, and sovereignty. Friction may likewise be required for trust, safety, accountability, and public order. The problem arises when these boundaries unnecessarily interrupt a Subject's legitimate realization of value or isolate elements that need to interoperate within a particular function or proprietary domain.

This condition is defined here as mis-correspondence in connection. The connection order required on the other side of that condition is defined as Universal Correspondence.

Universal Correspondence is a universal connection order in which the counterparties, information, functions, trust, authority, and execution elements required for a Subject's intent to be realized as value can be contextually corresponded around the Subject, without being structurally dependent on a particular function or proprietary domain.

Universal Correspondence therefore does not mean integrating everything into a single system or eliminating all differences. Individuals and organizations, domains and nations, different technologies and functions, and different systems of trust and governance must be able to retain their independence and sovereignty. Universal Correspondence does not mean Universal Integration. Its purpose is not to unify the interiors of different domains, but to enable relationships and actions to be established across their boundaries when needed while preserving their independence.

This correspondence has a clear point of reference. That point of reference is the Subject, not the Function.

This white paper abstracts the basic structure of connection through the relationship among Subject–Intent–Function. The Subject is the origin and point of attribution of value exchange. Intent expresses what the Subject seeks to achieve through connection. Function encompasses the information and functional resources mobilized to realize that intent. In this context, value exchange is not limited to monetary transactions. It includes diverse forms of value supplied, consumed, shared, or realized through connections among subjects, including information, knowledge, services, rights, relationships, and actions.

Universal Correspondence realigns the existing relationship. Instead of making the Subject subordinate to individual functions, the Subject becomes the persistent point of reference around which the counterparties, information, functions, trust, authority, and execution required by its intent are dynamically corresponded.

This transition does not imply dismantling or replacing the existing Internet or the World Wide Web. Rather, it builds upon the universal foundations for transmission and information connectivity that they have already established, while requiring a new logical upper structure capable of spanning existing function-centered domains and corresponding the elements necessary for a Subject's intent and value realization.

This white paper proposes HWW (Hyper-Web World Wide) as a Universal Correspondence Architecture for implementing this structure.

This white paper does not attempt to present a complete and final system of Universal Correspondence. It presents the core principles and architecture developed to date as a common foundation upon which technology, trust, governance, and other necessary domains can evolve independently while remaining capable of correspondence.

For Universal Correspondence to develop into a universal connection order in the real world, progress in a single technology or a limited set of domains will not be sufficient. Continued research, implementation, validation, and choice across different technologies, professional fields, institutions, and societies will be required.

This white paper therefore does not declare HWW complete. It presents an initial technical proposal for beginning its construction. Its further development remains open to the knowledge and efforts of those who will test, revise, and extend the architecture across diverse fields.

View contents

1. Toward Universal Correspondence
2. Chapter 1. HWW: Foundational Philosophy and the Architecture of Universal Correspondence
3. Chapter 2. HWW Architecture and Universal Correspondence Mechanisms
4. Part A. Core Architecture of HWW
5. Part B. HCTP and Dynamic Correspondence
6. Part C. Core Logical Modules and Trust–Authority Architecture
7. Part D. Correspondence Spaces, Governance, and Network Organization
8. Part E. Boundaries, Interoperability, and Federation
9. Part F. Recommended Reference Architecture
10. Chapter 3. HWW Operational Architecture and Interaction Flows
11. Part A. Operational Components and Interconnection Architecture
12. Part B. Formation and Preparation of the HWW Operational Basis
13. Part C. Core Correspondence and HCTP Operation
14. Part D. Composite, Federated, Governance, and Recovery Operations
15. Chapter 4. HWW Implementation, Operation, Applications, and the Correspondence Economy
16. Part A. Implementation and Operation
17. Part B. Use and Applications
18. Part C. Operational Economy
19. Part D. Subject-Centered Correspondence Economy — Changes in and Analysis of Economic Relationships
20. Part E. National Development and Expansion

HCTP Core Specification Draft 1.0

English 183 pp. · PDFPreface and contents open · full text by email

The draft specification of HCTP, the common correspondence grammar of HWW: common semantics, messages, operation, transport, and conformance. It is not a standard; it is a draft under technical validation.

Abstract and Status of This Document

Abstract

This document specifies the HWW HCTP (Hyper-Context Transaction Protocol) Core, the common correspondence grammar of HWW (Hyper-Web World Wide), a Subject-centric Universal Correspondence Architecture for the agentic AI era. HCTP Core defines the minimal common semantics, message architecture, operation model, transport binding, and cross-cutting requirements by which a Subject's Intent can be corresponded with counterparties, Functions, Trust, Authority, and Execution across heterogeneous domains, while each domain keeps its own internal structure. It treats AI and Agents as optional, delegated, and bounded capabilities rather than sources of Authority. HCTP Core does not replace the Internet, the Web, or existing identity, agent, and tool protocols; it defines the correspondence boundary at which they meet. Requirements are stated using BCP 14 key words.

Status of This Document

This document is Draft 1.0 of the HWW HCTP Core Specification. It is not a standard and has not been adopted by any standards body. It is based on the HWW White Paper and is published for review and implementation study. Open decisions and drafting status are listed in Section 13. Comments are welcome and may be sent to the author.

Contents

1. 1. Introduction
2. 2. Conventions, Definitions, Terminology
3. 3. Common Semantic and Reference Model
4. 4. Common Message Architecture
5. 5. Operation Model
6. 6. Transport and Binding Core
7. 7. Versioning, Extensions, Registries, Migration
8. 8. Cross-Cutting Normative Requirements
9. 9. Interface and Operation Families
10. 10. Conformance
11. 11. Profiles and External Bindings
12. 12. Security, Privacy, Governance, Operational Considerations
13. 13. Open Decisions and Drafting Status
14. 14. References
15. Appendix A. Recommended Reference Architecture (Informative)
16. Appendix B. Relationship to Existing Protocols, Identity Systems, and Agent Interoperability Mechanisms (Informative)
17. Appendix C. Illustrative Correspondence Patterns (Informative)

About Us

PAX S&T Co., Ltd. is a technology company developing HWW (Hyper-Web World Wide), a new internet architecture for the era of AI agents.

Today's internet is organized around functions, so the identities, data, and relationships of people and organizations are divided and confined service by service. PAX S&T has designed the Universal Correspondence architecture, which moves the reference point of connection from functions to Subjects. This architecture, which links external identifiers such as telephone numbers with internal identifiers such as DIDs through the Universal Key, is registered as Korean Patent No. 10-2990000, and an international patent application (PCT/KR2026/012600) has been filed.

HWW aims to be a public foundation of connection that everyone shares, not one company's platform. Individuals, companies, and nations connecting with the world while keeping control over their own data and connections, whatever AI they use: that is the new order of connection PAX S&T is opening. We publish the white paper and the specification, and we are looking for partners to realize it together.

All members of PAX S&T Co., Ltd.

Collaboration & Contact

HWW goes beyond one company's proprietary product: it is a new upper architecture for the internet that many Subjects, industries, and nations must build and develop together. Completing it requires knowledge of networks and security, identity and trust, AI, data, public policy, and international standards.

Areas of collaboration

• Technical validation and research: Reviewing the soundness, security, and implementability of the architecture and the HCTP specification
• Implementation and interoperability: Message, state, and authority processing, schemas, conformance testing, SDKs, and pilot environments
• Public infrastructure and policy: Applying identification, trust, and governance frameworks and common correspondence infrastructure for nations and institutions
• International cooperation and standardization: Interoperability discussions with international organizations, research institutes, and industry
• Services and ecosystem: Companies and developers who want to build AI agents and services on HWW

Contact
Please send inquiries about research, implementation, standardization, and pilot collaboration, and about our publications, by email. Email: jdookim0307@gmail.com