LLM Chat

Chatbot

To access the Chatbot click on Chatbot on the left navigation bar :

This interface allows you to ask questions to QAnswer chatbot. It's a Large Language Model (LLM) interface. If you choose to use QAnswer LLM, you will be asking questions to a secure chatbot, and your data will be safe.

• Initially, the chatbot is not linked to any document you provided on your AI Assistant section.
• You can choose another LLM by clicking on the LLM name in the bottom right of the question input.

Initial Chat

Data sources

Secure LLM

General LLM

Public LLM

Public Chat

info

These datasource settings work in exactly the same way as AI Assistant Data Source section!

You can ask questions about any topic you want.

Question

Answer

Edit Question

Conversation History

Conversation Actions

• Once you receive an answer you can regenerate it by clicking on Re-Generate answer.
• Additionally, you can also edit your question by clicking on appearing just below your question on hover.
• On the left panel you can see the history of each conversation you've started with the chatbot.

You have acces to several actions by clicking on the three dots

on the right of the chat title:

• Rename the conversation by clicking on .
• Share the conversation by clicking on .
• Delete a conversation you no longer use by clicking on .

Web Search

You can also activate "Search the Web" for specific needs by clicking on

on the left of the input field. It will toggle the search mode. If enabled, it will look like this .

AI Assistant

• You can ask a question to one of your AI Assistant by selecting it on the New Chat screen.
• You can also tag the AI Assistant in the chatbot by typing @ and its name in the input.
• If you want to remove the AI Assistant tag, you can click on the X icon next to the AI Assistant name in the text field.

Select AI Assistant

Question

Answer

@AI Assistant

Share Conversation

Sharing

To share a conversation click on

in actions menu, next to the conversation title.

• First, choose the type of sharing you want and its corresponding tab.
• You can share by Link and adjust access right to make it accessible to everyone that is logged in or not accessible.
• You can share directly on the app with users, a team or your organization if you are part of one.
• You can choose the name of users, teams and organizations you want to share your conversation with.
• You must choose the access type "Read only", "Read and Write" or "Use Only".
• Validate your choice by clicking on .
• You can update an access type by selecting another access type on the related dropdown.
• You can remove a share authorization by clicking on .

Share settings

Authorizations

Shared

Receiving

• You can only unsubscribe from a user-shared conversation. To do so click on an confirm.
• In read only access the user receiving the conversation will not be able to ask a question to the chatbot.
• In read and write access the user receiving the conversation will be able to ask a question to the chatbot and add datasource.
• In use only access (for chatbot conversation) is the same as read and write access. Use only mode difference is noticable for shared AI Assistants.

Result

Read Only

Read Write

Conversation Unsubscribe

RAG Mode

If you have provided a document that doesn't fit the window context in your chat conversation datasource, the chatbot will use RAG (Retrieval-Augmented Generation) mode to better understand and answer your question.

RAG mode

Modal Text

Why RAG Mode?

Your chatbot is currently using RAG (Retrieval-Augmented Generation) mode because your data doesn't fit entirely within the LLM's context window.

What does this mean?

The LLM can only see a portion of your data at once. In RAG mode, the system retrieves the most relevant chunks of information to answer your questions, but the LLM doesn't have visibility into all your data simultaneously.

How to fix this:

• Change the LLM to one with a larger context window
• Reduce your dataset size by removing unnecessary documents

info

RAG mode is not a limitation - it's a powerful technique used by most production AI systems. However, upgrading to a larger context LLM can improve responses for questions requiring broader context.

note

If you didn't create an AI Assistant yet you can check Create an AI Assistant section!