New Blue Oceans.

01.1 Measure the gap — without turning people into traffic

ITU estimates that in 2025, 6.0 billion people used the Internet while 2.2 billion remained offline. The GSMA’s 2026 connectivity report estimates 4.8 billion mobile Internet users at the end of 2025 and 3.4 billion people not using mobile Internet. It also reports that 38% of the global population live within mobile-broadband coverage without using mobile Internet. These measures have different definitions and overlapping populations: they must not be added together or presented as an addressable customer base.

Network coverage is not the same as an affordable device, usable information, accessible physical surroundings or the ability to complete a task. Someone may already be online and still be excluded by an inaccessible machine; another person may need a community hub or trusted human support rather than a personal smartphone.

01.2 Recognize unmet demand — without inventing orders

The United Nations’ 2026 SDG assessment finds that among 139 targets with trend data, only 36% are on track or making moderate progress, 49% are progressing too slowly and 15% have regressed. The UN also reports a 23.1% decline in official development assistance in 2025. These are indicators of unfinished development and financing pressure, not a statement that an HSIMC project has obtained funding.

UNDP’s SDG Investor Platform already researches investment opportunity areas by geography, sector and development priorities. HSIMC’s proposed additional contribution is an engineering method for moving from a validated need to a qualified local capability, a named payer or sponsor, an authorized action and an accepted result. Market intelligence is the start of diligence, not proof of revenue.

01.3 Start with one Reality Delta, not with a fashionable device

Every candidate begins with four questions: What can a person not yet do? What existing physical capability could help? Who has authority and responsibility for the action? Who will finance, deliver and accept the result? The answer may involve a repaired product, an accessible instruction, a shared workstation, a supervised human service or a local MCP-ready node — not necessarily a new app.

A cable carrying an NFC tag could open accessible product instructions and a service request on a compatible device. It does not itself provide network coverage, prove identity, authenticate a licensed product or authorize a physical machine. QR, a readable short URL, speech and trusted offline assistance must remain possible wherever they are needed.

01.4 Specify the Work before activating physical capability

A proposed HSIMC Human Capability Work record would identify the human objective, place and time, person or responsible organization, qualified local resources, required permissions, safety conditions, applicable language, cost and payer, evidence to be captured, acceptance criteria and continuity. A qualified local executor may decline or stop unsafe work; human authorization must be narrow, revocable and auditable.

MCP can help expose and invoke tools across systems, but a stateless request or tool description does not by itself create durable authorization, accountability, safety certification or a completed Work. These persistent responsibilities belong to the independently governed Reality, Authority, Work and Evidence layers. This remains an evolving HSIMC reference architecture, not a ratified international standard.

01.5 Connect SDG direction, ESG evidence and measurable outcomes

Relevant official targets include 4.4 and 4.6 (skills and literacy), 5.b (empowering technology), 8.3 and 8.5 (enterprise and decent work), 9.3 and 9.4 (small enterprise access and resource efficiency), 10.2 (inclusion), 11.2 (accessible transport), 12.5 (waste prevention, reuse and recycling), and 17.16–17.17 (multi-stakeholder partnerships). These are alignment references, not a claim that one product fulfils an entire SDG.

ESG assertions require different evidence: who actually took part, whether consent and accessibility were real, whether work was fairly paid, what material was genuinely reused, which safety checks passed, and the complete environmental and economic baseline. “Recycled”, “low-carbon”, “AI-enabled” and “inclusive” must never be treated as measured impact simply because they appear in a catalogue.

01.6 Keep public evidence separate from pilots and commercial acceptance

Three candidate paths illustrate what could be tested: an Accessible Product Passport linked to an existing Pannex accessory; human-assisted mobility and workplace participation informed by InnoSpire’s accessibility experience; and supervised repair or low-risk circular work supported by local engineering, including possible specialist partners such as 易联顺. The FIFA Women’s World Cup 2027 product concept, where relevant, is subject to a separate licensing and customer-approval chain.

Evidence levels remain distinct. Official UN, ITU, GSMA, UNDP and Hong Kong public materials establish context. Technical records and product catalogues establish only the capabilities or claims they actually document. Partner consent, paid contracts, field acceptance, real revenue, proven carbon savings and human outcomes must be evidenced separately. This Chapter reports no newly accepted global field pilot or contracted revenue arising from these proposed paths.

01.7 Make the economics and global invitation real

The person benefiting from a Work need not be the person paying. A buyer may fund an accessible product service; a transport operator may procure support; a manufacturer may pay a local repair partner; a research sponsor may fund a bounded experiment. Each relationship needs its own consent, contract, price, actual delivery cost, ongoing support and clear acceptance. SDG relevance alone produces neither a grant nor a purchase order.

The first meaningful milestone is modest and testable: one real barrier, one participant-controlled objective, one qualified local capability, one identified funding path, one safe field result and one human-accepted record. HSIMC publicly invites people to bring barriers, contribute existing capabilities, sponsor a bounded Work or offer qualified local execution. This is an invitation to discuss potential work, not an operational intake, financing portal or active MCP service.

02.1 1997: a documented standards contribution and its enduring lesson

The published foreword of GB 16844—1997, Self-ballasted lamps for general lighting services — Safety requirements (普通照明用自镇流灯的安全要求), names 廖炎 (Liao Yan) among multiple standard drafters. The national standards registry independently confirms its release on 3 June 1997 and implementation on 1 May 1998. This historical standard was superseded; it is not a presently valid compliance basis. The HSIMC founder identifies this named contributor as himself; the scan independently proves the printed attribution, while the link to the present individual remains founder-attested rather than supported by a separate identity certificate.

The standard drew upon an IEC international safety standard and included conditions for testing safety and interchangeability. Its relevance to HSIMC is a working method: independent manufacturers require shared technical language, specified boundaries and reproducible acceptance methods to cooperate reliably.

This experience does not imply that the contributor alone wrote or approved the national standard, nor that a lighting standard anticipated MCP. It motivates research into open capability contracts for physical AI, with new safety and field validation appropriate to each use case.

• Original evidence: GB 16844—1997 foreword, page 1, names 廖炎; page 3 records the 1997 approval date.
• Official status: issued in 1997, implemented in 1998, superseded in 2009; historical contribution, not a currently effective safety standard.
• Identity scope: named archival drafter plus founder’s personal confirmation; no separately issued personal credential or sole authorship claim.

02.2 From components to supply chains: cooperation has an economic reality

Over subsequent decades the same work expanded into physical products, power, cables, tooling, product engineering, quality checks and international supply chains. The experience is that an interface succeeds only when a supplier can manufacture it, a buyer can approve it, a user can operate it, and somebody can support it when conditions change.

Pannex and established manufacturing and customer relationships remain real, independent commercial assets. HSIMC neither replaces those relationships nor requires a whole factory or customer base to adopt one overarching architecture. A tested accessory, wiring interface, product identity or local service can contribute to a particular Work under terms that all parties accept.

The shift from product specifications to Capability Contracts therefore needs more than API documentation: quantities, tolerances, territory rights, acceptance tests, warranty, maintenance, worker compensation and cost allocation matter in the real world.

• A technology demonstration is not a customer order.
• An available supplier is not an automatically contracted implementation partner.

02.3 2026: a named Smart Traffic Fund project, research completed

The Hong Kong Smart Traffic Fund official project register identifies CS/57/2212/RA, “Developing an AI computer-vision solution to facilitate mobility for persons with visual impairment” (開發一套便利視障人士出行的人工智能電腦視覺解決方案). It names 創啟社會科技有限公司 as the applicant, Dr 蕭啟穎 as contact, a 12-month project period and HK$1,514,000 in approved funding. The public project status reads “研究已完成” — research completed. The funding figure records an approval, not independently verified cash received or commercial revenue.

InnoSpire’s Chinese-language website identifies itself as 創啟社會科技有限公司, providing a public match between its operating name and the applicant named in the project register. Separately, the Transport Department’s 13 April 2026 InnoEX announcement describes an AI smart-glasses project funded for mobility assistance, without naming InnoSpire or HSIMC in that bulletin. The project-specific government programme record, rather than a group photograph or similarity of descriptions, establishes the applicant and funded research identity.

Officially reporting “research completed” is a meaningful project milestone, not an independently supplied final acceptance certificate, verified road-safety test result, device certification, commercial deployment or endorsement of HSIMC. The public moment is therefore a catalyst for responsibility: can research results become independently validated, accessible and locally supported human capability under real operating conditions?

• Official programme identity: CS/57/2212/RA · applicant 創啟社會科技有限公司 · 12 months · approved HK$1,514,000.
• Public programme progress: 研究已完成 (research completed); formal final acceptance, field-safety certification and outcome audit have not been supplied.
• No implied endorsement: grant approval and public exhibition concern the named applicant/project, not the adoption of HSIMC’s GenAI Physical architecture.

02.4 Beyond a good AI answer: the barrier is often physical

InnoSpire notes an instructive limitation: a person with visual impairment may have access to text recognition on a phone and still be unable to aim the camera accurately at the text. Its glasses project addresses part of that physical interaction challenge using a wearable camera and spoken output. The important difference is between software that can recognize information and an arrangement that a person can reliably use.

This suggests a broader research direction from visual assistance to human capability. Accessible product information, voice or tactile interaction, human support, wearable observation and qualified local devices can be combined according to an individual’s choices. Capability must never be measured only by the model score or a staged demonstration.

Independent safe travel cannot be assumed from an image-recognition feature. Co-design with intended users, explicit consent, systematic testing of false negatives and environmental limits, optional human assistance and the right to stop are necessary before claiming a safety-critical outcome.

• User-defined outcome: a person can perform a previously difficult task under agreed conditions.
• Failure evidence: unsafe suggestions, inaccessible steps, unavailable devices and situations requiring a human handoff.

02.5 Stateless MCP communication; persistent human authority and evidence

The published MCP 2026-07-28 architecture has a stateless core: each protocol request carries its required protocol version and capabilities. The host coordinates clients, enforces consent and security boundaries, while servers expose focused tools and resources. This is useful for independent systems to communicate; MCP itself is not a license to operate physical equipment or a substitute for applicable product and safety standards.

HSIMC proposes an interoperable GenAI Physical reference architecture and Capability Contract profiles above such transport. A proposed profile would declare the capability owner, device and place, actual readiness, permitted scope, human supervisor, time limits, safety interlocks, evidence and acceptance authority. Stateful identity, grants, Work history, liability and audit trails must persist in governed systems, not disappear when one stateless request ends.

A local device may expose a verified capability through a narrowly authorized tool, then return a measured outcome or a refusal. No cloud model, MCP tool list, QR code or NFC tag may be assumed to confer authorization. These profiles remain HSIMC research candidates, not adopted ISO/IEC standards or externally approved certification schemes.

• Communication: independent and interoperable where technically supported.
• Responsibility: named humans and entities retain authority, obligations, stop control and recorded acceptance.

02.6 Economic sustainability turns responsibility into ongoing capability

A public mission without funding cannot be maintained globally. Manufacturers need approved orders, engineers and local operators need fair pay, research requires resources, and capital providers require transparent risks and a credible return or public-benefit mandate. These are legitimate requirements, not an embarrassment to an inclusive mission.

The person benefiting from an accessible service need not be the person paying. A purchaser, employer, public service operator, donor under a valid grant, local partner or product brand may fund different parts of the Work. Contribution, pricing, tax and licensing rights, procurement and support responsibilities must be separately agreed and documented.

The proposed work must show its actual financing conditions before launch: confirmed budget or price, eligible payer, direct and recurring costs, safety and warranty liabilities, who collects money, and how real outcomes will be accepted. SDG relevance and participation in a public exhibition cannot be counted as sales, grants, equity or cash received.

• No forced four-company venture; independent partner economics are negotiated by Work.
• No participant must prove commercial profitability to deserve accessible assistance.

02.7 From personal conviction to accountable global cooperation

The 1997 experience provides a standards-minded method. The 2026 Hong Kong public moment supplies a human reminder. The step still to be proven is operational: can an independent local user and an independent local provider jointly complete one useful task with clear authority, verified safety, fair payment where applicable and human acceptance?

HSIMC invites people to bring a real barrier, offer an existing capability, fund a bounded pilot or act as a qualified local executor. Start with a user-defined Reality Delta; identify what already exists, what is missing and who may authorize a small safe action. Then test, document results and failures, evaluate affordability and support, and only replicate what has been independently accepted.

The invitation applies to InnoSpire, Pannex, potential technology integrators including 易联顺, universities, local companies and other partners without presuming their agreement, identity rights or joint ownership. The goal is not another app ecosystem or a claim to direct the world; it is a repeatable contribution to human capability under real contracts and constraints.

• Bring One Real Barrier → Qualify One Capability → Fund One Work → Verify One Human-Accepted Outcome.
• Next evidence gate: one named use case with consent, partner rights, actual costs and an independently auditable result.