Search
What Is a Conversation?
A conversation in ADM is an interactive session between you and the AI system, similar to a continuous dialogue where:
- Context is maintained throughout the session
- Each message builds upon the previous one
- The AI remembers prior exchanges
- You can branch naturally into related topics
This design allows ADM to understand your intent, maintain continuity, and deliver more accurate responses.
Anatomy of A Conversation
Conversation Lifecycle
Each conversation follows a simple lifecycle:
| Stage | Description |
|---|---|
| Initiation | You start a new conversation or resume an existing one. ADM loads your context, preferences, and initial prompt. |
| Active | Messages are exchanged. ADM processes input, invokes agents, triggers workflows, and updates context. |
| Paused | The session is inactive but saved. Context is preserved, and the conversation appears in your history for later use. |
| Archived | Older conversations are searchable but not active. You can restore them as needed. |
Context and Memory
How ADM Maintains Context
ADM uses multiple layers of context to interpret and remember your interactions.
| Source | What It Tracks |
|---|---|
| Conversation History | Messages, previous questions, responses, and actions taken. |
| User Profile | Your role, preferences, and frequently used terms. |
| Knowledge Base | Organization-specific documents, uploaded files, and domain knowledge. |
| System State | Current data catalog, recent policy changes, and asset status. |
Context Limits
Each conversation has a context window that determines how much past information ADM can reference. For very long or unrelated discussions:
- ADM may summarize earlier parts of the session.
- Start a new conversation for unrelated topics to maintain clarity and performance.
Best Practices
Do's
- Start conversations with clear intent
- Provide context for ambiguous terms
- Use follow-up questions to refine responses
- Star important conversations for later reference
- Use descriptive titles when renaming conversations
Dont's
- Mix unrelated topics in same conversation
- Assume AI knows unstated context
- Use conversations for real-time monitoring
- Share sensitive information in conversation titles
Conversation History and Management
Accessing Conversation History
You can view and manage your past conversations to stay organized.
- Click the conversation icon in the left sidebar.
- The History Panel opens, displaying conversations grouped by:
- Last 7 days
- Last 30 days
- Older conversations
Filtering Options
| Filter | Description |
|---|---|
| All | Displays all your conversations. |
| Starred | Shows conversations you’ve marked as important. |
| Shared | Lists conversations shared with or by you. |
Conversation List View
Each conversation in the list shows:
- Title or first few words of the conversation
- Last activity timestamp
- Participant avatars (for shared items)
- Star indicator (if marked)
- More options (⋮) menu for management actions
Managing Conversations
From the More Options (⋮) menu, you can:
- Rename a conversation
- Star or unstar it
- Share it with specific team members
- Delete it permanently
- View detailed metadata
Was this page helpful?
On This Page
Core ConceptsWhat Is a Conversation?Anatomy of A ConversationConversation LifecycleContext and MemoryHow ADM Maintains ContextContext LimitsBest PracticesDo'sDont'sConversation History and ManagementAccessing Conversation HistoryFiltering OptionsConversation List ViewManaging Conversations