I make founders readableto AI systems
I'm Ekaterina Shalel, a founder and legibility strategist. I help founders and companies get read right by ChatGPT, Claude and Perplexity, with the Legibility Sprint, the public decision record, the Indifference Test and the First Customer Tax
Start here: how to become visible to AIWhat legibility actually means
People used to find you through search. Now they ask ChatGPT, Claude or Perplexity, and the model answers from whatever it can piece together about you. If that picture is thin, out of date, or tangled up with someone else's, you don't come up. Not because you're unknown. Because you're unreadable.
That's the difference I work on. Visibility is how often you get mentioned. Legibility is whether a model can tell exactly who you are, see how your companies and projects connect, and repeat a claim about you without guessing. Plenty of well-known people are illegible, and it costs them the recommendation.
The work itself is unglamorous: one sentence that describes you the same way everywhere, page code a machine can parse, a public record that ties your name to what you built, and a fixed set of questions you re-run later to see whether anything moved. Most people never do it, which is exactly why it works.
llms.txt A plain text file on your site that tells AI systems who you are, in their reading order rather than a human's.
Structured data (JSON-LD) Invisible labels in your page code. A human sees a paragraph; a machine sees "this is a person, this is her role, this is her company."
Wikidata The open database models check when they want a fact about a person or company confirmed by something other than your own website.
Test set A fixed list of questions you ask the models before and after the work, so a change is something you measured, not something you felt.
Terms I coined
The Legibility Sprint
Seven days to build the one place that tells the truth about you, and to make sure machines can read it. Day by day: your own site as the source, llms.txt, structured data, a Wikidata record, a scorecard, and a check on day 21 to see what changed
In plain terms: you stop hoping the models get you right and start giving them something correct to read.
The Legibility Audit
A documented snapshot of what GPT, Claude, Perplexity and Gemini say about you today, the sources behind those answers, and a prioritized 30-day plan. I ran the protocol on myself first; the Audit is me running it on you.
