How Spilled's smart features work

Spilled uses smart language technology in two specific ways:

1. Tagging your entries. When you write an entry, the text is sent to a third-party language service to identify what tags should apply (cancellation, effort, ghosted, broken promise, and so on). The service looks at the words you used and matches them to relevant tags. This happens almost instantly.

2. Writing personalized pattern reads (Pro only). When a pattern fires on your connection and you are a Pro subscriber, you can tap the pattern to get a personalized read. Spilled sends a summary of your recent entries to a third-party language service, which writes a paragraph or two of personalized analysis about what this pattern looks like in your specific situation.

What the smart features do not do:

• It does not read or store your entries for any purpose beyond tagging or writing the read you requested.
• They do not learn from your data. Your entries are not used to train any model.
• It does not produce diagnostic labels (it will not tell you he is a narcissist).
• It does not give you advice on what to do. Spilled is information, not therapy.

The third-party language service receives the text temporarily to process your request and returns the result. Spilled does not store the entry text on any server.