Using the assistant in a note
The AI assistant opens in a side inspector panel beside your active document. You can brainstorm outlines, draft explanations, check reasoning, or translate paragraphs directly within your workflow.
When you ask a question, the assistant incorporates the text of your active note as conversational context. You can copy generated markdown directly into your document or request specific edits.
What it sends and where
Tessamark operates no central AI proxy servers, collects no token metrics, and holds no log copies of your conversations. Gavware never sees your queries.
- Direct transmission: When you submit a prompt, Tessamark sends an HTTP request directly from your machine to the endpoint configured in your settings.
- Local execution: If you use a local model server (Camelid (Local), Ollama (Local) or LM Studio (Local)), your prompts and document context never leave your local machine. Network traffic is addressed entirely to
localhost.
- Cloud providers: If you use an external provider (OpenAI, OpenRouter, xAI, Google (Gemini) or Anthropic (via OpenRouter)), your request travels directly to that provider's public API using your personal API key.
Provider configuration
To configure your model endpoints, set up your personal keys, or switch between local daemons and cloud models, see the complete step-by-step instructions in the documentation:
→ AI assistant setup in documentation (the presets are Camelid (Local), Ollama (Local), LM Studio (Local), OpenAI, OpenRouter, xAI, Google (Gemini) and Anthropic (via OpenRouter)).
