I have spent much of my working life close to the point where an idea stops being a drawing and has to survive the physical world.
That experience includes standards work, machine vision, industrial projects, supplier coordination and the less glamorous work that comes after a promising concept: checking dimensions, finding the right connector, making a fixture, correcting a drawing, explaining the same requirement to different engineering teams, and discovering that a result that looked obvious in software is not obvious at all on the factory floor.
The arrival of powerful GenAI makes this gap more visible, not less visible.
A model can now produce code, drawings, analyses, translations and alternative designs at extraordinary speed. But the physical world still asks questions that cannot be skipped. What material is actually available? Which machine can process it? Who is allowed to use that machine? Which version is being built? Is the interface electrically compatible? Who accepts the result? What happens if the network disappears after a physical action has already started? Who receives the value when a task is completed?
I became convinced that one of the most important opportunities of the GenAI era is not simply to build more autonomous machines. It is to make existing capability easier to understand, combine, verify and reuse.
That capability is everywhere. It may be a sophisticated UWB team in Shenzhen. It may be a T736 computing module and a supplier’s development work. It may be an old printing shop in Kwun Tong that can make a precise mark or a jig before a distant engineering team has finished a meeting. It may be an embroidery machine that can place conductive thread. It may be a local worker who knows how to handle a difficult handover in a hotel. It may be a small workshop with a laser marker, a camera, a roll of copper foil, a stack of packaging material, a sewing machine, or one reliable way to make one thing repeatedly.
The challenge is not to pretend that all of these things are the same. The challenge is to give each contribution a clear boundary: what it can do, under what conditions, who has authority, how another team requests it, and what evidence proves the result.
This is why the white paper begins with stone and thread.
Stone represents material that already exists - sometimes valuable, sometimes discarded, sometimes poorly described. Thread represents familiar skill and ordinary machinery that can acquire a new function when combined with conductive materials, electronics and a defined interface. The credit-card-sized starting point represents a physical scale that people can immediately understand. It is small enough to hold, print, carry, attach and discuss, while the intelligence behind it can remain distributed across local surfaces, electronic modules, edge computing, existing equipment and cloud services.
The white paper then moves outward. It asks how people keep authority when models and tools become more capable; how MCP’s stateless protocol direction can support collaboration without erasing real-world state; how hardware and software teams can begin together before final hardware ships; how one function can be reproduced by multiple makers; how value should be measured without double counting; and how a successful reference result might mature into a shared specification without casually calling itself a standard.
The principle that holds these parts together is simple:
A defined result, at a named place and time, under agreed conditions, completed by a responsible party and verified by evidence.
I believe this matters economically, technically and socially.
Economically, it can shorten the distance between a GenAI plan and something a customer can use or buy. Technically, it lets different specialists work against the same boundary instead of waiting for one monolithic product team. Socially, it creates a path for people and small organizations to contribute useful functions without first becoming software platforms or fully automated factories.
That is the meaning of “No One Left Behind in the GenAI Physical Era” in this document. It is not a claim that technology automatically creates inclusion. It is a design requirement: participation should be understandable; authority should be visible; contribution should be attributable; evidence should travel with the result; and value should not disappear into a system that the contributor cannot inspect.
My role as Founder of HSIMC is not to claim that every piece is already solved. It is to help define the interfaces, bring complementary capability together, insist on evidence boundaries, and keep moving from plan to tested result.
The invitation is therefore deliberately practical:
Bring your existing capability.
Deliver one reproducible function.
Measure one useful outcome.
Make GenAI plans become real value - now, here,
globally. - Benny (Yan) Liao
Founder, HSIMC
12 November 2026