In the chat you have two kinds of features: tools that depend on the selected AI model (web search, connectors, document creation), and features that are always available: Magic Prompt, voice input and upload.
Model tools
Web search, image generation and document creation are up to the AI models: whether they are available depends on the capabilities of the selected model (shown in the model info card). The Smart Router automatically selects a model with the right capabilities.
The three main tools of a model are:
Web search: searches the internet for up-to-date information
Connectors: access connected external systems
Document creation: creates documents directly from the chat
Turning tools on and off
If you have selected a model manually, a tools icon appears. Use it to switch the main tools on or off as needed:
Deliberately disable web search, for example when the AI should work only with your own documents
Switch connectors on and off individually
Enable or disable document creation
This also works when the model has the capability in principle: you decide what gets used.
Magic Prompt
With Magic Prompt you improve your input with one click: enter your text, click the Magic Prompt button, and the AI automatically analyzes and expands your prompt.
Tip: Magic Prompt turns good prompts into better ones, but it does not replace a clear task description. Use it as a starting point and refine the result if needed.
Voice input and Voice Mode
You don't have to type everything:
Dictate: click the microphone icon and speak your text
Voice Mode: have a real conversation with the AI, completely hands-free: you speak your requests and the answers are read aloud to you. Ideal on the go or when you are working on something else at the same time
Tip: Choose the voice for speech output in your Profile Settings, four preset voices are available.
Upload
Use upload to give the AI material to work with:
Upload documents: drag files directly into the chat or select them
Screenshots: capture them directly from the chat
Knowledge bases: select one or more knowledge bases as context
Magic Prompt, voice input and upload work independently of the selected model.
Related articles
Questions? Message us right here in the chat.