-
Notifications
You must be signed in to change notification settings - Fork 0
Usage
Baldri edited this page Feb 11, 2026
·
1 revision
- Click New Chat (plus icon in the sidebar)
- Select a provider and model
- Type your message and press Enter
Responses are streamed in real time.
You can switch AI providers at any time:
- In chat: Click the provider name at the top of the chat window
- New chat: Choose a different provider when creating a conversation
- Default provider: Set in Settings under General
The sidebar lists all conversations chronologically:
- Click a conversation to open it
- Start a new conversation with the plus icon
- Conversations are saved automatically
Conversations are automatically titled based on their content. You can change the title at any time.
When the knowledge base is enabled:
- Ask questions about your documents
- Mingly automatically searches for relevant content
- Responses include source citations
Example:
"What does our project plan say about Q2 goals?"
Mingly finds relevant sections in your indexed documents and provides a grounded answer.
Mingly automatically tracks your API usage:
- Analytics tab in Settings shows costs per provider
- Token counter for each message
- Budget settings for monthly limits
- Open Settings > Budget
- Set a monthly limit
- Mingly warns you before the limit is reached
| Action | Shortcut |
|---|---|
| Send message | Enter |
| New line | Shift + Enter |
| New chat | Ctrl/Cmd + N |
| Settings | Ctrl/Cmd + , |
The more specific your prompt, the better the response:
- Good: "Explain the benefits of TypeScript over JavaScript for large projects"
- Better: "List 5 specific benefits of TypeScript over JavaScript for a team of 10 developers"
Mingly remembers the conversation history. You can reference previous messages:
"Explain the third point in more detail"
- Quick questions: Haiku / GPT-3.5 (affordable, fast)
- Complex tasks: Claude Opus / GPT-4o (precise, detailed)
- Local use: Ollama (free, no cloud)
Go to Settings > General and add at least one API key.
- Check your internet connection
- Make sure the API key is valid
- Some providers have regional restrictions
- Check if your API plan supports the model
- Some models require separate access approval
Next: FAQ
User Guide (EN)
Benutzerhandbuch (DE)
Developer