Zum Hauptinhalt springen

Setting how AI works – Standard or Advanced

Under “Settings > Integrations > AI provider” you choose “How AI works” in the “AI configuration” form: “Standard” works with the current document only, “Advanced” also checks suggestions against your team’s data such as text modules, at the cost of more tokens and longer run times.

Last updated:
Video transcript
In this video, you decide in preeco | datenschutz how much context the AI uses for its suggestions. In the settings, expand “Integrations” and select “AI provider”. Under “How AI works” you see the current setting. Click “Edit”. The “AI configuration” panel opens. “Standard” balances result quality, token usage and run time. The AI knows only the document it works on. “Advanced” also checks suggestions against your team’s data, such as your text modules. It is more precise but uses more tokens and takes longer. The “Generation mode”, by contrast, controls how thoroughly the model reasons. You can combine the two freely. Tick “Consent to data transmission” and click “Save”. The application then tests the connection to the provider. The setting applies to the whole team. A failed connection test is listed under “AI activities” with the provider’s message. Tip: with a curated text module library, “Advanced” pays off. The AI reuses your approved wording instead of writing similar text again.

The “How AI works” setting decides how much context the AI uses for its suggestions. In “Advanced” mode it also checks its suggestions against your team’s existing data, such as your text modules, and links matching modules instead of writing copies of them.

Prerequisites

  • You are signed in as an administrator or may edit team management.
  • An AI provider with a valid API key is set up, or you set it up in the same step.

The two modes

  • “Standard”: a balance of result quality, token usage and run time. The AI knows the document it works on and nothing else.
  • “Advanced”: the AI takes more information into account. Results are more precise, token usage is higher and suggestions take longer.

Choose the mode

  1. Click “Settings” in the left sidebar, expand “Integrations” and select “AI provider”.
  2. The “AI provider” section shows the current setting under “How AI works”. Click “Edit”. The “AI configuration” panel opens.
  3. Under “How AI works”, select “Standard” or “Advanced”.
  4. Check that “Consent to data transmission” is ticked.
  5. Click “Save”. The application tests the connection to the provider.

The setting applies to the whole team and to all AI features. A failed connection test is listed in the “AI activities” section with the function “Connection test” and the provider’s message.

How it differs from the generation mode

The “Generation mode” in the same form controls how thoroughly the model reasons. “How AI works” controls which information it gets to do so. You can combine the two independently.

Practical tip

If your team keeps a curated text module library, “Advanced” pays off: the AI reuses your approved wording instead of writing similar text again, and the library stays free of duplicates. For quick one-off suggestions in a small team, “Standard” is enough.

Changes and errors may occur. The information in this article has been carefully compiled, but does not claim to be complete or correct.