This article provides an overview of how conversations work in the AI Assistant and the core capabilities available to employees and administrators.
Overview
A conversation is a persistent chat session between an employee and the AI Assistant. It serves as the container for all messages within a single thread.
Conversations are:
- Persistent - Saved automatically and available for later access
- User-scoped - Visible only to the individual user
- Multi-channel - Available across supported platforms through a shared backend
This model allows users to move between devices and channels while maintaining continuity.
Supported channels
The conversation framework supports multiple client surfaces, including:
- Web and mobile applications
- Collaboration platforms such as Microsoft Teams
- Microsoft Outlook
Additional messaging and assistant platforms may also be supported, depending on deployment.
Conversation lifecycle overview
Conversation creation
Conversations begin when an employee starts a new chat experience.
Automatic titles
The system generates a conversation title based on the initial interaction. Titles help employees recognize conversations in history and can be edited later.
Ongoing continuity
Conversations support multi-turn exchanges and remain available for future reference. Employees can return to prior conversations at any time.
Conversation history
Employees can browse prior conversations through a dedicated history panel.
Key characteristics:
- Conversations are grouped by recency (for example, Today, Yesterday, This week, Older)
- Entries display summary information such as title, preview text, and timestamp
- Items are ordered by most recent activity
- History loads progressively as users scroll
Conversation discovery
Search
Employees can search conversation titles within the history panel to quickly locate prior discussions.
Direct access
Conversations can also be opened through direct links, enabling deep linking and browser navigation support.
Conversation organization
Employees can manage conversation metadata to keep history organized.
Capabilities include:
- Renaming conversations for clarity
- Deleting conversations when permitted by organizational policy
Administrative permissions may control availability of deletion features.
Conversation starters
Conversation starters help employees begin new chats quickly.
Key characteristics:
- Displayed when starting a new conversation
- Selecting a starter begins a new conversation with a prefilled prompt
- Starters may vary based on application context (for example, page or product area)
URL-based conversation routing
URL parameters can guide how conversations open and start.
Supported behaviors include:
- Opening a specific conversation
- Prefilling a message without sending
- Prefilling and automatically sending a message
- Creating a new conversation when one is not specified
These capabilities support deep linking, automation, and embedded scenarios.
Channel-based conversation behavior
Conversation handling varies by platform.
Web and portal experiences
Employees explicitly create and select conversations. Each new chat creates a distinct conversation thread.
Messaging and assistant platforms
Some channels maintain a single ongoing thread for a defined session window.
Examples include:
- Microsoft Teams
Session duration and thread controls depend on platform capabilities and organizational configuration.
Administrative considerations
Access control
Organizations define:
- Which employees can delete conversations
- Availability of conversation management capabilities
Data lifecycle
Conversation deletion permanently removes:
- The conversation container
- All associated messages
Retention and compliance requirements should be evaluated before enabling deletion features.
Cross-channel consistency
Conversation state is unified at the backend, ensuring continuity across supported channels.