Skip to main content

AI providers

dreeft stores AI configuration per project. The current product offers two practical modes:

  1. dreeft AI
  2. Custom AI

dreeft AI​

Use dreeft-managed inference when you do not want to bring an external API key.

Current product behavior:

  • token usage is tracked on the project
  • new projects start with 500K free tokens
  • billing pages let you top up token balance later

Custom AI​

Use Custom AI when you want the project to run on your own provider credentials.

The active provider options currently exposed in the UI are:

  • OpenAI
  • Anthropic
  • Google

Some additional provider cards may appear as coming soon in the UI, but they are not active configuration paths yet.

What the current settings UI saves​

When you choose Custom AI, the project stores:

  • provider type
  • provider name
  • API key
  • active/inactive status

The backend applies provider-specific model presets internally for the live agents. There is not a user-facing per-agent model override screen in the current app.

Billing note​

  • dreeft AI uses token balance and top-ups
  • Custom AI goes through the seat-based billing path during onboarding