Ansrom

Example page: Ayşe Yılmaz is a fictional creator, made up to show how Ansrom works. The content, numbers and answers are samples.

Digital marketing

AY

Ayşe YılmazAI twin

Search, retrieval and the craft of writing for machines

Ask questions, get answers in Ayşe Yılmaz’s own words and reasoning, with the source for every claim. This is an AI trained on published content, not Ayşe Yılmaz in person.

  • Publishing since 2021
  • Content updated Oct 1, 2026
  • 204 items

What it knows

I write about how published content gets found, indexed and quoted: by search engines, by language models and by the people who read both. Since 2021 I have run a weekly newsletter on retrieval and content structure, a video series where I test ideas on my own archive, and a set of long essays on product decisions in search.

  • 142The Retrieval LetterNewsletter
  • 38Retrieval, liveVideo
  • 24EssaysEssay

The twin answers only from this content. When a question falls outside it, it says so instead of guessing. Your messages are never used to train the twin.

A sample answer

A sample exchange with the twin. Sources sit beside the answer; tap a marker to see where a claim comes from.

Why do you recommend chunking by section instead of fixed-size windows?

AI twin· Ayşe Yılmaz

Because section boundaries carry the structure I already put into the piece. When I write a newsletter, each heading groups one claim with its evidence, so splitting there keeps the two together . Fixed windows cut straight through that: you get a chunk holding the claim and a neighbour holding the reason, and retrieval only ever sees one of them.

I tested both on my own archive and the section split won on every question that needed a justification rather than a bare fact . Short sections are fine; anything under about 120 words I fold into the next chunk as overlap .

Packages

Every package includes cited answers in your language and the AI-twin disclosure. Choose how much depth and memory you want.

  1. Starter

    $14per monthLaunch price · $29 per month after launch

    300 messages a month

    • Answers with sources
    • Answers in your language
    • Remembers the conversation
    Subscribe

  2. StandardRecommended

    $29per monthLaunch price · $59 per month after launch

    450 messages a month

    • Answers with sources
    • Answers in your language
    • Deeper, more thorough answers
    • Remembers you across conversations
    Subscribe

  3. Advanced

    $59per monthLaunch price · $119 per month after launch

    600 messages a month

    • Answers with sources
    • Answers in your language
    • The most thorough answers
    • Longest memory across conversations
    Subscribe
  • Billed monthly. Cancel at any time; the subscription ends at the end of the paid period.
  • Prices exclude VAT where it applies.
  • Subscribers must be 18 or older.

Questions

From $14 per monthSee packages