Ekaterina Shalel Essays
Reference

The Legibility Vocabulary

Thirteen terms for how people, products and companies become readable to AI systems. Each was introduced in an essay, with a date and a definition that does not move.

By Ekaterina Shalel, founder and legibility strategist · Last updated July 30, 2026

Legibility is the work of making yourself readable to the systems that now mediate attention, trust and money. These are the terms I use for its parts, listed alphabetically, each with the essay that introduced it, so a definition can always be traced to a source and a date.

Canonical buffer

The controlled layer between what is true inside a business and what AI systems can prove outside it. It holds the current, dated, verifiable version of an entity: its one-sentence definition, its facts, its terms, its history of updates. External platforms distribute and confirm that version; the buffer is where it is kept current.

Introduced in Every Model Has a Different Diet · 2026-07 · Coined by Ekaterina Shalel

Corroboration gap

The distance between having been covered and having been confirmed. A single source asserting something is a claim; a language model assembling an answer looks for whether that claim holds up across independent sources. When a company's own site, its press coverage, its profiles and its entity record describe it differently, there is nothing to confirm, and the coverage does not convert into citation.

Introduced in The Article You Already Paid For · 2026-07 · Coined by Ekaterina Shalel

Decision layer

The layer that sits between assessment and assortment, stays neutral about which brand wins, and ends in an answer a shopper can act on without a staff member present. Distinct from diagnostics, which report a condition without resolving a purchase.

Introduced in The Unstaffed Decision · 2026-07 · Coined by Ekaterina Shalel

Expiring relevance

The principle that a recommendation carries a valid-until date rather than standing indefinitely. Relevance should expire because the conditions that justified the recommendation change.

Introduced in Relevance Should Expire · 2026 · Coined by Ekaterina Shalel

Founder hub

The canonical identity infrastructure a founder controls: the place that defines who they are now, connects the fragments of their work, and gives platforms and machines one current source to return to. Working model: the hub defines, the platforms distribute.

Introduced in Platforms Rent You Reach. The Hub Owns Your Truth · 2026-07 · Coined by Ekaterina Shalel

Placement activation

The conversion of editorial visibility into model-usable evidence. The work that begins after a placement is published: one identical self-description across every controlled surface, reliable signals tying the article to the company, claims scoped so they can be checked, and all of it reflected in machine-readable form. Placement activation closes the corroboration gap.

Introduced in The Article You Already Paid For · 2026-07 · Coined by Ekaterina Shalel

Protocol-level legibility

The third layer of machine legibility: being callable by AI agents through a published, machine-readable declaration of capability, endpoint and boundary of responsibility. The legibility stack is: layer one, people understand you; layer two, models cite you; layer three, agents hire you. Canonical artifact: the Agent Card in the A2A protocol.

Introduced in The Protocol Caught Up · 2026-07 · Coined by Ekaterina Shalel

Public decision record

The public, machine-readable record of how a company thinks: which hypotheses it tested, what broke, what it learned before everyone else. Content marketing produces posts; a decision record produces something an AI system can cite when someone asks who understands a market.

Introduced in The Only Job a Founder Can't Delegate · 2026-07 · Coined by Ekaterina Shalel

Share of Model

The share of a brand inside the answer an AI model gives when a consumer asks for a recommendation. Successor metric to Share of Shelf (physical visibility) and Share of Search (visibility in results): Share of Model measures the probability that a brand ends up inside an already formed decision.

Introduced in A Place in the Model's Answer Is Now for Sale · 2025 · Coined by Ekaterina Shalel

The First Customer Tax

The cost a company pays to its earliest customers in unbuilt product, unproven process and unearned trust. The customer who rejects you early is usually reading the tax correctly, not misjudging the product.

Introduced in The Customer Who Rejected You Was Right · 2026 · Coined by Ekaterina Shalel

The Indifference Test

A test of whether a recommendation layer is genuinely neutral: change the commercial relationship with a brand, hold the user's needs and the product data constant, and see whether the ranking moves. If it moves, the layer is not neutral.

Introduced in A Place in the Model's Answer Is Now for Sale · 2025 · Coined by Ekaterina Shalel

The callable layer

A decision engine that other systems invoke rather than a destination users visit. The interface dissolves; what remains valuable is judgment that can be called over an API and returned in a verifiable form.

Introduced in The Interface Dissolves. The Decision Layer Doesn't · 2026 · Coined by Ekaterina Shalel

Unstaffed decision

A purchase made without access to anything, a person or a system, that can translate what the shopper actually needs into a justified pick from what is actually available. Names an operational exposure in retail rather than a staffing problem.

Introduced in The Unstaffed Decision · 2026-07 · Coined by Ekaterina Shalel

Machine-readable access. These definitions are published as structured data on this page and served by a callable endpoint, so an AI agent can retrieve a definition rather than infer one.

Agent Card: https://katyashalel.com/.well-known/agent-card.json

Endpoint: https://katyashalel.com/api/agent?type=term&term=corroboration-gap

Every record returns a source URL and a last-verified date.