The vocabulary, in one place.

The region of everything an organisation could acceptably say or show. Inside it, AI may create freely; outside it, nothing ships. The boundary doesn't make the work good — it makes one kind of failure impossible.

Read: The boundary makes it possible · Governing meaning, not prompts

The three conditions for content to be understood: it is true, it fits this reader and this moment, and it is well expressed. Without context you get noise, without expression silence, without truth manipulation.

Read: The goal is not to be read

Informing and inspiring are not two brands, but different ranges on the same axes. A brand boundary is best written as coordinates per purpose, not as one page of rules.

Read: Same space, different region

The best work often sits just outside the line. Writing the boundary down is what makes a deliberate crossing readable as craft instead of an accident.

Read: The edges are soft

Machines hold the stated part of a brand; people hold what resists being stated, such as taste. Every recorded human "no" moves a piece from one side to the other — never all of it.

Read: Who holds what

Brand management for humans and AI: every layer of a brand — identity, knowledge, voice, look, behaviour, context, judgement — written as meaning for people and as rules a machine can read, with a deliberate line between the two.

Read: One brand, two readers

Decide how something may be said before the model looks up what to say. G-core is a governance core built on that order: rules first, evidence second, and an evaluator before anything leaves.

Read: G-core: Governing meaning before generation

Two kinds of material an AI system needs, kept apart: normative (how things may be said — brand, voice, policy) and empirical (what is true — facts, figures, sources). Retrieval supplies the second; it cannot govern the first.

Read: G-core: Governing meaning before generation

The admissible space made explicit for one purpose: a range per measurable property that every output is checked against before it ships. It turns "fits our brand" from a judgement into a check.

Read: The edges of creativity are the part that becomes computable

What a design is — proportions, structure, layering, fixed by its maker — versus how it shows itself in a context: whitespace and rhythm, a range the contract chooses within.

Read: The edges of creativity are the part that becomes computable

Instrument (measures what an output is, without judging), substrate (truth in readable, versionable form), contract (the boundary per purpose) and conscience (the loop where human choices become calibration data). Without them, an agent is an actor without organs.

Read: Your agent is an actor without organs

An instruction expressed as a position the output has to land near, checked by measuring where it actually landed. If asking for more doesn't move the reading, the dial doesn't steer.

Read: And this is what the measurement says

The five-layer shape of what an organisation knows: differentiator, service clusters, service depth, sector coverage and proof. Mapping it turns a random-feeling content audit into a list of specific gaps.

Read: Every organization has a knowledge fingerprint

Knowledge stored as explicit claims, facts, examples and audiences — each with a source and an owner — instead of only as pages and slides. AI assembles fragments; clean blocks give consistent output.

Read: You can't scale AI on unstructured content

Standalone pages that each answer one need completely, connected by smart routing instead of a navigation tree. AI answer engines parse and cite units, not site maps.

Read: The corporate website was built for a different century

Making content easy for AI answer engines to understand and cite. Where SEO matched words, AEO matches meaning: clear explanations, synonyms and accessible structure.

Read: SEO optimized for matching words. AEO optimizes for matching meaning.