Asking Questions

Edgebric answers your questions using the documents in your data sources. It searches your documents, finds relevant passages, and generates an answer with citations — all running locally on your Mac.

How It Works

When you ask a question, Edgebric:

1. Searches your documents using both semantic search (understanding meaning) and keyword search (exact matches)
2. Ranks results using Reciprocal Rank Fusion — a technique that combines both search methods for better results
3. Generates an answer using the AI model, with the relevant document passages as context
4. Cites its sources — each answer includes references to the specific documents, sections, and page numbers used

This is called Retrieval-Augmented Generation (RAG). The AI model doesn't make things up from thin air — it bases its answers on your actual documents.

Asking a Question

1. Open the chat interface from the main page
2. Type your question in natural language — just ask like you would ask a colleague
3. Press Enter or click Send
4. The answer streams in as it's generated

Good questions:

• "What is our parental leave policy?"
• "How do I file an expense report?"
• "What were the key decisions from the Q3 planning meeting?"

Tips:

• Be specific. "What are the PTO rules for contractors?" works better than "Tell me about PTO."
• If you don't get a good answer, try rephrasing or ask a more targeted question.
• You can scope your query to specific data sources using the source picker in the chat interface.

Citations

Every answer includes citations showing where the information came from:

• Document name — Which file contains the source material
• Section — The heading or section within the document
• Page number — For PDFs, the specific page

Click a citation to view the original document passage. This lets you verify the answer and read more context.

If Edgebric can't find relevant information in your documents, it will tell you rather than guessing.

Answer Types

Edgebric classifies each answer:

Type	Meaning
Grounded	Answer is fully based on your documents
Blended	Answer combines document information with general knowledge
General	No relevant documents found; answer uses the model's general knowledge
Blocked	Query was filtered for safety reasons

Conversations

Edgebric remembers context within a conversation. Follow-up questions reference earlier messages automatically:

You: What is our vacation policy?

Edgebric: [answers with citations]

You: How does that apply to contractors?

Edgebric: [answers in context of the vacation policy discussion]

Managing Conversations

• Conversations are listed in the sidebar
• Click New Chat to start a fresh conversation
• Archive or delete old conversations from the conversation list
• Archived conversations can be restored later

Converting to Group Chat

If a conversation would benefit from collaboration, you can convert it to a group chat:

1. Open the conversation
2. Click the options menu
3. Select Convert to Group Chat
4. Invite team members and share data sources

See Group Chats for more on collaborative querying.

Privacy Modes

When asking questions, your privacy mode affects what gets recorded:

Mode	What's recorded
Standard	Conversation history stored for future reference
Private	Queries are anonymous — not linked to your identity
Vault	Everything stays on-device, encrypted

See Privacy Modes for details on each mode.

File & Image Attachments

If your active model supports it, you can attach files or images to your query:

• Images — For models with vision capabilities, attach a screenshot or photo and ask questions about it
• Documents — Attach a document directly in the chat to ask questions about its content without adding it to a data source

Look for the paperclip icon in the chat input area. It appears when your active model supports attachments.

Feedback

Each answer has thumbs up/down buttons. Your feedback helps track answer quality. If an answer is wrong or unhelpful, thumbs down it and optionally leave a comment explaining what was wrong.