Skip to main content

AI Chat

The AI Chat page is the first entry under App Channels. It allows users to ask free-form questions and receive AI-generated answers using one or more knowledge source configurations (KSCs).

Before You Start

Before using AI Chat, ensure you have:

Key Capabilities​

Channel/KSC Selector​

A dropdown at the top of the chat interface that determines which knowledge sources the AI can access for responses:

How It Works:

  • Each channel is linked to a specific Knowledge Source Configuration (KSC)
  • Switching channels changes the underlying knowledge sources and available content
  • Different channels may have different confidentiality levels and content access
  • The AI's knowledge and response style may vary based on the selected channel

Selection Impact:

  • Knowledge Scope: Different channels access different knowledge repositories
  • Content Restrictions: Confidential channels access restricted content not available in public channels
  • Response Quality: Channel configuration affects the AI's ability to provide accurate, relevant answers

Confidentiality Rules: Channel determines which KSC is used; KSC determines which sources can be retrieved. See: Confidentiality rules

Prompt Suggestions and Quick Actions​

The chat interface provides several helpful features to get started and manage conversations:

Prompt Suggestion Cards:

  • Common Queries: Preset suggestion cards display helpful example questions like "What are your products?"
  • Getting Started: Sample prompts help new users understand the AI's capabilities
  • Context-Specific: Suggestions are tailored to the selected channel's knowledge sources
  • Quick Launch: Click any suggestion card to automatically submit that question

Conversation Management:

  • Pin Conversation: Use the pin icon to save important conversations for easy access later
  • Star Messages: Add valuable AI responses to your knowledge base for future reference
  • New Conversation: Start fresh conversations while preserving your chat history
  • Add to Knowledge: Contribute information back to the knowledge sources when you identify gaps

Conversation Area​

The central interface displaying the complete dialogue between you and the AI:

Message Display:

  • User Messages: Your questions and inputs displayed on one side
  • AI Responses: Generated answers with source attribution and confidence indicators
  • Timestamps: When each message was sent and received
  • Message Threading: Clear conversation flow with proper context preservation

Progress Indicator:

  • Generation Status: Shows when the AI is actively generating a response
  • Processing Time: Indicates how long the AI is taking to respond
  • Cancellation Option: Ability to cancel response generation if taking too long
  • Source Retrieval: May show when AI is searching knowledge sources

Conversation History:

  • Persistent Memory: AI maintains context throughout the conversation
  • Reference Capability: AI can reference earlier parts of the conversation
  • Context Building: Understanding improves as the conversation progresses
  • Clear History: Option to start fresh while maintaining logged conversation data

Message Composer​

The input interface at the bottom of the chat for submitting questions and managing conversations:

Input Options:

  • Text Input: Type questions directly into the text box
  • File Upload: Upload documents, PDFs, and other files for the AI to analyze
  • Image Upload: Upload images that the AI can process and analyze
  • Send Methods: Press Enter or click the send button to submit
  • Multi-line Support: Shift+Enter for line breaks without sending

Quick Actions:

  • Pin Conversation: Save important conversations for easy access
  • Star Messages: Add valuable responses to your knowledge base
  • Clear History: Start a fresh conversation while preserving logged data
  • Export Chat: Download conversation transcript for sharing or reference

Input Enhancement:

  • Auto-complete: Suggestions based on common questions and previous queries
  • Spell Check: Basic spelling correction for better AI understanding
  • Drag & Drop: Drag files directly into the composer for upload
  • File Preview: See uploaded files before sending

How It Works​

Response Generation Process​

  1. Query Processing: AI analyzes your question for intent and key concepts
  2. Context Application: System applies conversation history and company context
  3. Knowledge Retrieval: AI searches relevant sources from the selected KSC
  4. Response Synthesis: AI generates a comprehensive answer based on retrieved information
  5. Source Attribution: Response includes references to knowledge sources used
  6. Quality Assessment: System evaluates response quality and flags potential issues

Conversation Management​

  • Context Preservation: AI maintains conversation context across multiple exchanges
  • Follow-up Awareness: Subsequent questions benefit from previous conversation context
  • Memory Management: System balances context retention with performance
  • Session Continuity: Conversations persist across browser sessions when logged in

Data Flow and Logging​

  • Response Logging: All interactions are logged for performance analysis
  • Privacy Controls: User data handling follows configured privacy settings
  • Feedback Integration: User feedback helps improve future response quality
  • Performance Tracking: Conversations contribute to overall AI performance metrics
Important Features
  • Channel Selection: Choosing the right channel ensures access to appropriate knowledge sources
  • Prompt Suggestions: Use suggestion cards to understand the AI's capabilities and get started quickly
  • File Uploads: Upload documents and images to enhance your questions with additional context
  • Conversation Management: Pin important chats and star valuable messages for future reference
  • Knowledge Contribution: Help improve the AI by adding missing information when you find gaps

Troubleshooting​

No Answer or Wrong Answer​

  • Check Channel: Verify that the correct channel/dataset is selected
  • Knowledge Gap: If the AI cannot find relevant information, it might return a generic response
  • Add Knowledge: Use the "Add to Knowledge" feature to contribute missing information

Chat Not Loading​

  • Refresh: Try refreshing the page
  • Switch Dataset: Try selecting a different dataset
  • Browser Issues: Clear cache or try a different browser

Response Quality Issues​

  • Dataset Content: Ensure your knowledge sources contain the information you're asking about
  • Question Clarity: Try rephrasing your question to be more specific
  • Context: Provide more context in follow-up questions

Best Practices​

Before Starting a Conversation​

  1. Channel Selection: Always verify you've selected the appropriate channel/dataset for your needs
  2. Prompt Exploration: New users should try suggested prompt cards to understand AI capabilities
  3. Goal Clarity: Have a clear objective for what you want to accomplish in the conversation
  4. File Preparation: Gather any relevant documents or images you might want to upload

During Conversations​

  1. Specific Questions: More specific questions yield better, more targeted answers
  2. Context Building: Provide relevant background information for complex topics
  3. Follow-up Questions: Use conversation flow to ask clarifying questions and dive deeper
  4. Progress Monitoring: Watch the progress indicator and cancel if responses take too long
  5. Source Verification: Review source attributions to understand where information comes from

Managing Conversations​

  1. Pin Important Chats: Use the pin feature to save valuable conversations for easy access
  2. Star Messages: Add useful AI responses to your knowledge base for future reference
  3. Clear When Needed: Start fresh conversations when switching topics significantly
  4. Export Transcripts: Download important conversations for sharing or documentation
  5. Provide Feedback: Rate responses to help improve AI performance
  6. Contribute Knowledge: Use "Add to Knowledge" when you identify information gaps

Optimizing AI Performance​

  1. Conversation Length: Longer conversations often produce better, more contextual responses
  2. Question Refinement: Rephrase questions if initial responses aren't quite right
  3. Context Addition: Add more context if AI responses seem generic or off-target
  4. Multiple Angles: Ask about the same topic from different perspectives for comprehensive understanding
  • Knowledge Sources - Manage the information repositories that power AI responses
  • Channels - Configure which knowledge sources are available to different chat channels
  • Responses - Review and analyze all AI chat interactions