Skip to content
Saylikeby Unplain

Company

Privacy and data handling

Saylike is built to keep as little as possible. This page says exactly what it reads, what it sends where, and what it keeps.

Updated

The short version

  • No account, no sign-up, no email address.
  • Your uploaded files and the text extracted from them are processed in memory and dropped when the request ends.
  • Your finished skill is stored only in your own browser, so you can reopen it. “Forget it” on the homepage deletes it.
  • The writing is sent to one AI provider for the analysis (below). Nothing is sold or shared for advertising.

What happens when you generate a skill

  1. Saylike fetches your homepage and up to four more public pages of your site, plus its robots.txt and sitemap.
  2. Uploaded files are read in memory. Text files, PDFs and DOCX files are converted to text on our server; screenshots are sent to the AI provider to transcribe the text in them.
  3. The extracted text (up to about 24,000 tokens) is sent to the AI provider, currently Anthropic (Claude), to analyse the voice. Anthropic processes it under its commercial API terms, which do not allow training on API data by default.
  4. The analysis comes back, is checked against your text, and is turned into the skill, which is sent to your browser.

Nothing from these steps is written to a database or to disk. If you use “Write a sample”, your current skill text is sent to the same provider to write it.

What we keep

  • A short-lived cache for public websites. When a skill is generated from a website alone (no uploads), the result is kept in server memory for up to six hours so the same site asked for again doesn't cost a second analysis. It contains only what was public on the site. Results that include uploads are never cached.
  • Operational logs. For each generation we log the site's domain, how many pages and samples were read, word and token counts, the model used, the cost and how long it took. No file names, no text and no results.
  • Rate limiting. Your IP address is held in memory for up to an hour to limit how many skills one visitor can generate. It is not logged.
  • Anonymous usage counts. When analytics is switched on, it records what happens (a skill was generated, an export was used) without cookies and without the URLs, files or text you entered.

The Saylike crawler

Saylike fetches pages only when a person asks it to analyse a site, and at most eight pages per request. It identifies itself as SaylikeBot/1.0 in the user agent, follows robots.txt for every page except the homepage the person asked for, never runs JavaScript, never submits forms, and never visits addresses on private networks. If you'd rather it didn't read your site, add User-agent: SaylikeBot with Disallow: / to your robots.txt, or contact us.

Your content, your skill

You own the skill you generate. Please only analyse websites and documents you are allowed to use; a skill based on someone else's writing is still their writing.

Contact

Saylike is made by Unplain. Questions or requests: hello@unplain.io.