Documentation Index
Fetch the complete documentation index at: https://docs.sagalegal.io/llms.txt
Use this file to discover all available pages before exploring further.
Chat is where you work with AI in Saga. Ask questions about your documents, draft content, analyze materials, or run research — all through a natural language conversation. The AI draws on whatever context you give it and cites its sources so you can verify every answer.
Starting a conversation
Click Chat in the sidebar to open a new conversation. Type your question and press Enter to send.
You can also start a chat from inside a project. When you do, the AI automatically has access to that project’s knowledge base — every active document becomes context for the conversation without you needing to attach anything.
How context works
The AI’s answers depend on what information it has access to. In a project chat, that context comes from the project’s knowledge base. In a global chat (outside any project), you provide context by attaching files, selecting organizational sources, or enabling web search. You can also combine multiple sources in a single conversation.
For a full breakdown of how to control what the AI sees, see Context and sources.
Citations
When the AI references a document, it includes inline citations — small numbered chips linked to the specific passage it drew from. Hover over a citation to see the original excerpt. Click it to open the source document in the side panel, scrolled to the relevant section.
Citations work for project documents, legal research results, and web sources. They’re how you verify what the AI is telling you without hunting through files yourself.
Message actions
Hover over any message to see its available actions:
- Copy the message content to your clipboard
- Regenerate a new response to the same question
- Export to Word to download the message as a .docx file
- Edit your original message to get a revised response
- Like / Dislike to give feedback on the response quality
- Save to Prompt Library to reuse a prompt you’ve written
- Delete the message from the conversation
Message versions
When you edit a message or regenerate a response, Saga creates a new version instead of overwriting the original. Navigation arrows appear so you can move between versions.
This is useful when you want to explore different angles on the same question — for example, asking the AI to analyze a clause under two different jurisdictions, or rephrasing a question to see whether a broader or narrower framing surfaces better results. Each version is preserved, so you can always go back.
Tasks
Tasks are built-in commands for common operations. Access them from the Tasks button in the chat toolbar or type / in the input:
| Task | What it does |
|---|
/summarize | Generate a concise summary of the provided text or documents |
/anonymize | Redact personal identifiers and sensitive information |
/proofread | Check text for grammar, spelling, and style improvements |
/compare | Compare two texts or document sections and highlight differences |
Improve prompt
Before sending a message, click the pen icon next to the send button to let the AI refine your prompt. This can help you get more targeted results, especially for complex questions. If you prefer your original text, click the revert arrow to undo the improvement.
Click the microphone icon in the chat input to dictate your message instead of typing. Saga transcribes what you say into the input box, where you can edit it before sending. Recording is capped at thirty minutes per session and requires microphone access in your browser.
Voice transcription is powered by Deepgram. The language Deepgram listens for is tied to your Saga interface language — set it under Settings > Personalization > Default language. Supported languages are English, Norwegian, Dutch, Danish, Polish, Spanish, German, Italian, Portuguese, Turkish, French, and Swedish. If the UI is set to a language outside this list, Saga falls back to English transcription.
If you dictate in Polish but Saga’s interface is in English, the transcript will be garbled — Deepgram is listening for the wrong language. Switch the UI language to match the language you’re speaking.
Thinking mode
Some models show their step-by-step reasoning in an expandable section before the final answer. This is useful for complex legal questions where you want to see how the AI arrived at its conclusion. If you don’t need the reasoning, click Skip thinking for a faster answer to get a direct response.
Export a conversation
To download an entire chat as a Word document, click the export button in the chat header. The export includes all messages, citations, and the conversation title. You can also export individual messages from the message action menu.
AI-generated documents
When the AI drafts a document — such as a memo, summary, or contract clause — it appears as a chip below the response. Click the chip to open the document in the side panel. Hover over it and click the download icon to export it as a Word file.
Chat history
Your conversations are saved in the chat history, accessible from the sidebar. You can search chats by title, star favorites for quick access, rename conversations, or delete them. Chat history is organized with favorites first, then by most recent activity.
Chats are scoped to the account and workspace that created them. If a chat seems to be missing, check that you’re signed in to the same workspace (firms with multiple Saga subdomains have separate chat lists per subdomain), and remember that chats started inside a project live in the project — not in the global chat list. See Troubleshooting for more recovery steps.
How long chats are handled
You don’t need to manage the AI’s context window manually. The AI fetches document content on demand through tool calls rather than holding every document in its prompt, so long conversations no longer hit a hard percentage cutoff and there’s no “context full” warning in current versions of Saga.
Very long chats can still feel slower because the AI has more history to read. If you’re starting a new line of analysis or moving to a different document set, open a fresh chat — it’s faster and keeps the focus tight. For ongoing work over days or weeks, upload your documents to a project’s knowledge base instead of re-attaching them in each new chat.
