Free guide

Visibility Without Budget

Advertising buys attention that ends with the budget. Here's a system where people and AI systems find, read and cite you, and it compounds instead of burning out

Why it matters: search has shifted. Answers increasingly come from models, not result pages, and a model cites whoever it can read unambiguously. The job is no longer clicks, it's becoming a legible entity. This is called GEO, and it costs no budget, only discipline of wording. Tested on myself: top 20 authors on vc.ru and citability inside models without one ad dollar.

People no longer scroll results, they ask models. A model answers with whoever it can read unambiguously. First step: one canonical definition of who you are, identical everywhere, from the Instagram bio to llms.txt. Inconsistent wording tells the machine there are several of you, or none.

Prompt · canonical entity

Assemble a canonical entity definition from my data: [name, role, product, one line on what I do]. Give three versions of the same wording: 60 characters for a bio, 160 for a meta description, 300 for profiles. In a separate list, give every spelling of my name and brand, including transliterations, that should appear next to the definition. The wording must be precise with zero surrounding context.

Everyone owns the industry's common words, only you own yours. A term you coin, define and repeat consistently across platforms becomes a route: when a model explains that concept, it arrives at your texts. This is how I seeded the Indifference Test, expiring relevance and the callable layer, without a single ad dollar.

Prompt · your terms

Here are my texts: [paste two or three]. Find the ideas I phrase differently from the industry default. For each, propose a one-to-three word term, a one-sentence definition, and a check: is the term already taken by an established meaning. Discard anything that sounds like marketing, keep only what describes a mechanism.

One idea, several platforms, terms verbatim. Synonyms kill the seeding: if one text says callable layer and another says invocable tier, the machine sees two different concepts. Every version links back to the canonical source, so the weight accumulates in one place instead of smearing.

Prompt · one idea, three platforms

Here's an essay: [paste]. Adapt it for three platforms: [for example LinkedIn, Substack, an industry outlet]. Keep my terms verbatim, never substitute synonyms. For each version: its own opening two lines for that platform's audience, length native to the format, and a closing link to the canonical source of the text.

A site for machines is not design, it's statements: llms.txt with definitions, JSON-LD with a Person type, a canonical tag on every page, all name spellings side by side. This very site went from invisible to machine-readable in one day, and the path is reproducible.

Prompt · llms.txt

Assemble an llms.txt for my site. Inputs: [who I am, the product, my terms with definitions, links to canonical texts and profiles]. Write in short statements with zero marketing, so a model can quote lines directly. Structure: who, what I do, key concepts with definitions, where to read, official profiles.

Once a month, ask five different models: who is [your name], what is [your term]. The answers show where you're already legible and where it's empty. The empty spots are next month's publishing plan. Reach expires in the feed, legibility compounds.

Go deeper

This is guide two of four. The full course runs on a waitlist

First cohort · the waitlist closes August 6

Visibility Without Budget

Assemble your entity

Coin your terms

Publish where machines read

Make your site machine-readable

Measure legibility, not reach

Go deeper