Overview
Document Chat in Kodexa uses artificial intelligence to help you interact with documents through natural language conversations. Instead of manually searching through documents, you can ask questions and receive relevant answers based on the content of documents in your workspace.
What is Document Chat?
Document Chat is an AI-powered feature that:
Understands natural language questions
Searches across documents in your workspace
Provides relevant answers with source references
Learns from your interactions to improve responses
Supports follow-up questions for deeper exploration
Accessing Document Chat
To use Document Chat:
Open a Project workspace
Navigate to the Chat panel
Start a new chat or select an existing one
Starting a New Chat
To begin a new conversation:
Click the Start New Chat button
A new chat session is created
Type your question in the chat interface
The AI processes your question and responds with relevant information
Each chat session maintains its conversation history, allowing you to ask follow-up questions that build on previous responses.
Viewing Existing Chats
The Chat panel displays a list of your previous conversations:
Chats are sorted by creation date (newest first)
Click any chat to view its message history
The list automatically refreshes to show new messages
You can have multiple chat sessions for different topics
Using the Chat Interface
Asking Questions
Type natural language questions about your documents:
"What is the total amount on invoice 12345?"
"Show me all contracts expiring this month"
"Summarize the key terms in the agreement"
"Find all documents mentioning customer ABC Corp"
Message Templates
Pre-configured message templates provide quick access to common queries:
Click a template button to send a pre-written message
Templates are configured based on your document types
Save time with frequently used queries
Follow-Up Questions
The AI maintains context throughout the conversation:
Ask clarifying questions about previous answers
Refine your search based on initial results
Build on information from earlier in the conversation
How Document Chat Works
Document Access
Document Chat can access:
All documents in the current workspace
Extracted data from data definitions
Document metadata (filenames, dates, etc.)
Content from processed documents
AI Processing
When you ask a question:
The AI analyzes your question to understand intent
Searches relevant documents in the workspace
Extracts pertinent information
Formulates a natural language response
Provides source references when applicable
Chat Management
Creating Chats
New chats are created with default settings:
Named "New Chat" initially
Linked to the current workspace
Include configured AI participants
Can be renamed for better organization
Deleting Chats
You can delete chat sessions you no longer need:
Removes the chat and all its messages
Does not affect documents or data
Cannot be undone
Auto-Refresh
The chat list automatically refreshes every second to show:
New messages in existing chats
Newly created chats
Updates from AI responses
Best Practices
Be Specific - Clear, detailed questions get better answers
Use Context - Reference specific documents, dates, or data elements
Follow Up - Refine your query if the first answer isn't quite right
Organize Chats - Create separate chats for different topics or projects
Review Sources - Check the source references to verify information
Common Use Cases
Data Extraction
"What is the invoice date on document X?"
"List all line items from invoice 12345"
"Show me the vendor information from these receipts"
Document Search
"Find all contracts with Company ABC"
"Show documents uploaded last week"
"Which invoices are over $10,000?"
Summarization
"Summarize the key points of this contract"
"What are the main terms in this agreement?"
"Give me an overview of this report"
Comparison
"Compare invoice 123 and invoice 456"
"What's different between these two contracts?"
"Show me pricing changes across these documents"
Tips
The chat interface works best with documents already processed in the workspace
Create a new chat for each distinct topic or research area
Use natural language - you don't need special syntax
The AI learns from the conversation, so follow-up questions are often more accurate
Message templates save time for repetitive queries
Chat sessions are saved automatically as you interact