When the AI world realised it needed standards

Two years ago, one consultant saw what was coming

In early 2024, Tom Cranstoun attended CMS Kickoff, a Boye & Company conference for content management practitioners, and wrote an article for CMS Critic that described a problem almost no one was yet talking about.

AI, he argued, was not primarily a content creation tool. Its real impact would be as a content consumer. Machines were becoming web users, and the web was not designed for them. He described AI agents as behaving like eight-year-olds shopping for toys: reading simply, directly, ignoring brands, reviews, dropdowns, and marketing copy. He warned that websites put visual presentation above semantic structure, burying important information in ways that machines cannot reliably extract. He proposed a solution: design content for four device types: mobile, tablet, desktop, and machine.

That article was the starting gun.

Over the following two years, Tom developed those observations into a complete discipline: Machine Experience (MX), with a formal metadata layer, a governance model, and an architectural framework. He wrote two books. He built tooling. He created an open standard.

Then, on 10 March 2026, a small group of digital leaders gathered in Vancouver, no stage, no slides, no sales pitches, and confirmed everything he had been building towards.

Vancouver: the world catches up

The Boye & Company CMS Experts session was intentionally intimate, and as a result, the conversation became honest very quickly.

What emerged in that room was not a Canadian story. The same conversation is happening worldwide. Across continents and sectors, teams are facing the same accelerating pressures:

• AI adoption without governance.
• Content systems that cannot keep up.
• Leadership pushing for innovation.
• Operational teams trying to prevent chaos.
• Public-sector teams under-resourced but over-expected.
• Rising concerns about digital sovereignty.

Everyone is asking the same questions. No one has the framework.

These are exactly the problems Tom identified in 2024, and they have only intensified. The difference now is that he has the answer: The Gathering, an open, global standards body dedicated to Machine Experience.

A web outrunning its own structure

AI is moving faster than teams can adapt

In the Vancouver session, leaders described the same tension felt worldwide: AI tools are improving rapidly, but governance and risk frameworks have not kept pace.

The current global pattern looks like this: AI assists. Humans review. Humans approve. It is safe. It is necessary. It is not a long-term system.

Content has become infrastructure, but systems still treat it like pages

Participants explored the "content supply chain", where content is no longer a static web asset but structured information flowing across channels, platforms, and AI-assisted workflows. Most companies still operate with fragmented content models, legacy CMS assumptions, no unified metadata layer, and no machine-readable governance. This is a global bottleneck.

Digital sovereignty is now a strategic priority

The Vancouver discussion touched on Canadian sovereignty, but the underlying concerns are universal: where is data hosted? Who controls the platform? How do we secure long-term independence? These are strategic questions now, no longer technical ones.

Public-sector teams everywhere face the same pressure

Municipalities are being asked to deliver enterprise-grade digital services with small teams and limited budgets. This imbalance is global.

What's missing: a standard for Machine Experience

The Vancouver gathering confirmed what Tom wrote in 2024: the world is having honest conversations, but conversations alone will not fix structural gaps.

Back then, he called for an "AI Evangelist" role to bridge design, development, and content teams. Two years on, the need has grown well beyond a single role. What is missing is a framework that is open, global, machine-inclusive, community-governed, stable, versioned, and designed for both humans and AI systems.

This is what Tom spent two years building.

Where MX fits: Content Ops and the wider content estate

Content Ops is the discipline of creating, managing, improving, publishing, distributing, archiving, and retiring content across every digital channel. Machine Experience is the layer that keeps that work usable when an AI agent, or any other system, encounters the file outside the environment that produced it.

A website is a fraction of any team's content estate. Contracts, policy documents, product specifications, technical reports, briefings, and field manuals never reach the web, but AI agents are reading them anyway, inferring what they can and guessing the rest. Being on the web and being machine-readable are not the same thing.

MX is the DNA a file carries when it leaves any system. It governs what survives extraction, so the next reader can answer the same questions the originator used to answer.

MX builds on what teams already have. Schema.org and JSON-LD describe what entities mean. WCAG defines accessibility. Open Graph handles sharing. MX adds governance, provenance, lifecycle state, and agent affordances where those standards leave gaps. It never duplicates what they already cover.

The machine universe is expanding. AI agents, robots, autonomous vehicles, industrial systems, IoT devices, medical instruments, and classes of machine not yet invented all read the same files. Designing for that audience is the next move Content Ops has to make, and MX is the layer that lets it.

This is the four-part pitch in one breath:

• MX is the contract.
• The Gathering sets the standards.
• REGINALD does the signing.
• CogNovaMX operates the service.

The Vancouver room described the symptoms. Content Ops needs an infrastructure layer for the machine reader, and MX is that layer.

What Machine Experience actually is

Machine Experience is a discipline, not a product: a set of principles and patterns for making the web work for everyone and everything that uses it.

The core idea is deceptively simple. Design for both audiences simultaneously, not one at the expense of the other. When meaning is made explicit rather than implicit, it helps everyone. When information is structured semantically rather than just visually, it serves all users regardless of their capabilities or access methods. The same patterns that work for machine comprehension simultaneously improve human accessibility and human comprehension. Convergence, not trade-off.

MX is built on principles that emerged from two years of practical implementation.

Design for both means every design decision works for human developers and AI agents simultaneously. The best solutions serve both audiences without compromise.

Metadata-driven architecture means structured metadata makes content maximally machine-readable while remaining human-readable, operating at every level from repository down to individual code blocks.

Context declaration means files explicitly state what context they provide and what context they require. AI agents can read dependencies directly. Errors from incomplete understanding are prevented before they happen.

Universal accessibility means content is reachable by all types of AI agents regardless of their capabilities. Plain text formats, explicit markup, and declared relationships serve both disabled users and automated systems. WCAG requirements for disabled users provide proven patterns that also help machine readability. Accessibility and machine experience turn out to be the same design problem.

Executable documentation means documents carry their own generation instructions. The result is self-documenting specifications with executable build logic: one source of truth that is documentation and implementation guide at once.

Size-neutral documentation means avoiding hard-coded counts and using descriptive language that remains accurate as collections grow or shrink. This sounds minor until documentation drift silently undermines every automated system that reads it.

The result is a framework that operates at four layers, repository, directory, file, and code, with clear metadata at every level declaring purpose, audience, stability, and relationships. Every MX-compliant file carries enough context for an AI agent to read it without guessing. When agents guess, they hallucinate. When they have explicit structure, they do not.

The benefits run in both directions. AI agents get complete context, clear permissions, reduced errors, and faster navigation. Humans get self-documenting code, clear structure, reduced onboarding time, and better tooling.

These are not theoretical principles. They are the working architecture behind The Gathering's open standard.

The Gathering: open standards for Machine Experience

The Gathering is an independent, community-governed standards body dedicated to building open standards for Machine Experience.

It exists so that content, identity, and policy can be read, interpreted, and processed consistently by both people and AI systems. Where the Boye session revealed the global problem, The Gathering offers the global response.

A public, open drafting process

Standards evolve in the open. Anyone can participate. Stability comes through community consensus.

A universal metadata layer

Content becomes structured, portable, machine-readable, and interoperable across systems. This is what the Vancouver group was describing without yet naming.

Machine-inclusive governance

Instead of "AI assists, humans approve", The Gathering defines machine identity, machine-readable policy, provenance, accountability, and participation rules. It turns improvisation into infrastructure.

A constitutional community

The Gathering is not a discussion group. It is a governed ecosystem with roles, processes, a Code of Conduct, stable releases, and clear pathways for new work.

A gift to the world

This is the part that matters most. The Gathering is an open standard, created by Tom Cranstoun and offered freely to the world as a public good, neither proprietary nor a product. It is the structured response to the unstructured reality he first described in 2024, and that Vancouver confirmed the world is now ready to address.

Why this moment matters

The Boye & Company gathering captured a global truth: AI, content, governance, and sovereignty are converging faster than institutions can respond.

The world does not need more conversations. It needs standards, metadata, governance, operational clarity, machine-inclusive rules, open processes, and community stewardship.

It needs The Gathering.

Publishing schedule

The Gathering provides the open standard. The books provide the implementation guide. Two companion volumes translate MX from concept into practice:

The Handbook arrives first because the world needs to start building now. The Protocols follow because the world will need the complete reference once it does.

Closing thought

In 2024, Tom Cranstoun saw that machines were becoming users of the web, and that the web was not ready. In 2026, Vancouver proved him right.

Between those two moments, he built the standard, wrote the books, and created The Gathering.

This is the bridge between honest conversations and durable, global, machine-inclusive infrastructure, and it exists because he spent two years building it and gave it to the world.