Overview
Flow ID: persona-management
Category: Chat Interactions
Estimated Duration: 5-10 minutes
User Role: All Users
Complexity: Complex
Purpose: This flow enables users to configure multiple AI "personas" within a single chat conversation. Each persona can have different behaviors, contexts, and purposes - for example, one persona that searches datasets, another that provides creative responses, or multiple personas simulating a team discussion. This allows for sophisticated multi-agent conversations and specialized AI behaviors.
Related Flows
- Create New Chat from Template - Templates with pre-configured personas
- Chat with Dataset Query Enabled - Dataset context in personas
- Conduct Multi-Turn Conversation - Using personas in conversation
- Configure Chat Behavior Settings - Related settings
- Create New Empty Chat - Starting point for persona configuration
Prerequisites
Before starting, users must have:
- Application installed and running
- Active chat conversation (new or existing)
- Understanding that personas affect AI behavior
- Optionally: Dataset uploaded if planning to use dataset-query personas
- Patience to configure advanced settings
Step-by-Step Flow
Main Path (Happy Path)
Step 1: Open Active Chat
- User Action: Navigate to or create a chat conversation
- System Response: Chat interface displays
- UI Elements Visible:
- Chat conversation area
- Message input at bottom
- Settings gear icon (typically near input or in header)
- Visual Cues: Gear icon indicates settings available
Step 2: Access Advanced Settings
- User Action: Click the settings gear icon (typically in the input area or chat header)
- System Response: Advanced settings panel expands or opens
- UI Elements Visible:
- Expandable/collapsible settings panel (may slide in from side or expand inline)
- Multiple sections or tabs:
- General settings
- Prompt configuration
- Personas section (target area)
- Close or collapse control
- Visual Cues:
- Panel slides in with animation
- Sections clearly labeled
- Personas section stands out
Step 3: Locate Personas Section
- User Action: Scroll to or click on "Personas" section in advanced settings
- System Response: Personas section displays or expands
- UI Elements Visible:
- "Personas" section header
- List of current personas (if any exist)
- Each persona shown as a card with:
- Persona name
- Brief description or status
- Drag handle (for reordering)
- Edit and delete icons
- "Add Persona" button (prominent, likely blue)
- "Teed Up Personas" section (if any queued for manual execution)
- Visual Cues:
- Persona cards visually distinct
- Drag handles indicate reorderability
- Add button stands out
Step 4: Review Default Persona
- User Action: Observe the default persona (typically "Sophie - Assistant" or similar)
- System Response: Default persona is displayed
- UI Elements Visible:
- First persona card showing default assistant
- Name: e.g., "Sophie (Assistant)"
- Configuration summary visible or collapsed
- Edit and delete options available
- Visual Cues: Default persona may have special styling or icon
Step 5: Initiate Add New Persona
- User Action: Click "Add Persona" button
- System Response: Add persona modal or form appears
- UI Elements Visible:
- Modal dialog overlaying chat
- Title: "Add Persona" or "Create New Persona"
- Form fields:
- Persona Name (text input)
- Persona Prompt/Instructions (large text area)
- Context Type (dropdown: None, Static Text, Dataset Query)
- Context Placement (dropdown: Insert, Prepend, Append)
- Additional options (checkboxes):
- Disable generation (dataset results only)
- First person voice
- Disabled (inactive)
- "Cancel" and "Save" or "Add" buttons
- Visual Cues:
- Modal centered and prominent
- Required fields may have asterisks
- Form organized logically
Step 6: Name the Persona
- User Action: Type a descriptive name in the Persona Name field
- System Response: Text appears as typed
- UI Elements Visible:
- Text input with entered name
- Character count (if limited)
- Visual Cues: Active text field
- Example Names: "Legal Reviewer", "Dataset Expert", "Creative Writer", "Marketing Perspective"
Step 7: Write Persona Instructions
- User Action: In the prompt/instructions text area, write instructions that define this persona's role and behavior
- System Response: Text appears as typed, text area may expand
- UI Elements Visible:
- Large text area with instructions
- May have helper text or examples
- Character/token count possibly shown
- Visual Cues: Expandable text area
- Example Instructions:
- "You are a legal expert reviewing documents for compliance issues. Focus on risk assessment and regulatory concerns."
- "You provide creative, innovative ideas without constraints. Think outside the box."
- "You are a marketing professional analyzing content for audience appeal and messaging effectiveness."
Step 8: Select Context Type
- User Action: Click Context Type dropdown and select appropriate option
- System Response: Dropdown expands showing three options
- UI Elements Visible:
- Dropdown with options:
- None: No additional context (AI relies only on prompt and conversation)
- Static Text: Add custom text to every message
- Dataset Query: Automatically search dataset and include results
- Selected option highlighted
- Additional fields may appear based on selection
- Dropdown with options:
- Visual Cues:
- Dropdown clearly shows current selection
- Selection affects which additional fields appear
Step 9a: Configure Static Text Context (if selected)
- User Action: If "Static Text" was selected, enter text in the Context Info field that appears
- System Response: Text area for static context becomes available
- UI Elements Visible:
- "Context Info" text area
- Instructions explaining this text will be included with every message
- Visual Cues: Conditional field appears based on Context Type selection
- Use Case: Company policies, reference information, guidelines that should always be included
Step 9b: Configure Dataset Query Context (if selected)
- User Action: If "Dataset Query" was selected, system may auto-use active dataset or show dataset selector
- System Response: Dataset query will be performed automatically for this persona
- UI Elements Visible:
- Indication that dataset queries will be performed
- May show which dataset will be queried
- Number of results to retrieve (may be configurable here or in general settings)
- Visual Cues: Dataset icon or indicator
- Use Case: Personas that should always reference your documents
Step 10: Select Context Placement
- User Action: Choose how context should be included relative to user's message
- System Response: Dropdown shows three options
- UI Elements Visible:
- Context Placement dropdown with:
- Insert: Add as separate message before AI response
- Prepend: Add to beginning of user's message
- Append: Add to end of user's message
- Selected option highlighted
- Context Placement dropdown with:
- Visual Cues: Dropdown clearly indicates selection
- Note: Affects how AI interprets context
Step 11: Configure Additional Options
- User Action: Check or uncheck optional checkboxes based on desired behavior
- System Response: Checkboxes toggle on/off
- UI Elements Visible:
- Disable Generation checkbox: If checked, persona returns only dataset results without AI processing
- First Person checkbox: If checked, AI responds in first-person voice
- Disabled checkbox: If checked, persona exists but won't be used
- Visual Cues: Checked boxes have checkmarks, clear labels explain each option
Step 12: Save New Persona
- User Action: Click "Save" or "Add" button
- System Response:
- Modal closes
- New persona appears in personas list
- Settings are saved
- UI Elements Visible:
- Personas section now shows new persona card
- New persona appears in list (may be at bottom or top)
- Persona card shows name and brief configuration summary
- Visual Cues:
- Smooth modal dismissal
- New persona card appears with animation
Step 13: Review Personas List
- User Action: Look at the updated personas list
- System Response: All personas displayed
- UI Elements Visible:
- Multiple persona cards (default + newly added)
- Each showing:
- Name
- Drag handle (six dots or bars)
- Edit button (pencil icon)
- Delete button (trash or X icon)
- Configuration hints (icons for context type, etc.)
- Visual Cues:
- Cards clearly separated
- Drag handles indicate reorderability
Step 14: Optional - Reorder Personas
- User Action: Click and drag a persona card to reorder
- System Response:
- Persona card follows cursor
- Other personas shift to make space
- Drop persona in new position
- UI Elements Visible:
- Persona being dragged appears elevated/highlighted
- Drop zones indicated between other personas
- Real-time reordering preview
- Visual Cues:
- Drag and drop animation
- Visual feedback during drag
- Note: Order affects response sequence if multiple personas used
Step 15: Optional - Edit Existing Persona
- User Action: Click edit button (pencil icon) on a persona card
- System Response: Edit modal opens with persona's current settings pre-filled
- UI Elements Visible:
- Same modal as "Add Persona" but with "Edit Persona" title
- All fields populated with current values
- Can modify any field
- "Cancel" and "Save Changes" buttons
- Visual Cues: Pre-filled form indicates editing mode
Step 16: Close Advanced Settings
- User Action: Click close button or collapse settings panel
- System Response: Settings panel closes or collapses
- UI Elements Visible:
- Chat interface returns to normal view
- Conversation area and input fully visible
- Settings gear icon available to reopen
- Visual Cues: Smooth collapse/close animation
Step 17: Use Personas in Conversation
- User Action: Send a message in the chat
- System Response:
- If multiple personas configured, system may use first persona OR prompt for selection
- Persona's instructions, context type, and settings affect response
- Response reflects persona configuration
- UI Elements Visible:
- AI response attributed to specific persona name
- Response content reflects persona's instructions and context
- May see context information if configured
- Visual Cues:
- Persona name appears with response
- Response style matches persona configuration
Final Step: Personas Configured and Active
- Success Indicator:
- Multiple personas exist in settings
- Can edit, reorder, or delete personas
- Personas affect conversation behavior as intended
- Context is applied appropriately
- Responses match persona instructions
- System State Change:
- Chat has multiple persona configurations saved
- Conversation uses persona settings for responses
- Settings persist for this chat
- Can return to modify personas anytime
- Next Possible Actions:
- Continue conversation using configured personas
- Test different personas by modifying settings
- Create more personas for additional perspectives
- Manually trigger specific personas (if "teed up" feature used)
- Export or reuse persona configurations in new chats
Alternative Paths & Strategies
Strategy A: Quick Edit Default Persona Only
When to use: User wants simple customization without multiple personas
Steps:
- Open advanced settings
- Click edit on default persona
- Modify instructions to change AI behavior
- Save
- Use chat with modified single persona
Strategy B: Use Template with Pre-Configured Personas
When to use: Starting chat with template that includes multiple personas
Steps:
- Create new chat, select template (e.g., "Legal Review Team")
- Template automatically loads multiple personas
- Review pre-configured personas in settings
- Optionally adjust persona settings
- Use conversation with full team immediately
Strategy C: Manual Persona Execution ("Teed Up")
When to use: Want to control which persona responds to each message
Steps:
- Configure multiple personas
- In personas section, use "Teed Up" controls
- Queue specific persona(s) for next response
- Send message
- Only teed-up persona responds
- Repeat for different perspectives
Strategy D: Delete All and Start Fresh
When to use: Persona configuration became too complex, want to reset
Steps:
- Open advanced settings > Personas
- Delete each persona (except cannot delete all, must keep one)
- Edit remaining persona to be new default
- Add new personas with simpler configuration
- Save and use refreshed setup
Error States & Recovery
Error 1: No Persona Name Provided
Cause: User tries to save persona without entering name
User Experience:
- Error message: "Persona name is required"
- Cannot save until name entered
- Field may have red border
Recovery Steps:
- Enter a name in the Persona Name field
- Click Save again
- Persona should save successfully
QA Note: Form validation should prevent saving. If save button is disabled until name entered, error won't occur.
Error 2: Persona Instructions Empty
Cause: User provides persona name but no instructions
User Experience:
- Persona may save but behave generically
- No clear guidance for AI behavior
- May see warning about empty instructions
Recovery Steps:
- Edit persona
- Add clear, specific instructions
- Save changes
- Test persona to verify behavior improved
QA Note: Empty instructions not technically an error, but results in poor persona behavior. System may allow it.
Error 3: Context Type Mismatch
Cause: "Dataset Query" selected but no dataset available or activated
User Experience:
- Persona may show error during use
- Message: "Dataset not available"
- Persona fails to function properly
Recovery Steps:
- Ensure dataset is uploaded and activated
- Edit persona to verify Context Type settings
- OR change Context Type to "None" or "Static Text" if dataset not needed
- Save and retry
Error 4: Cannot Delete Last Persona
Cause: User tries to delete all personas (system requires at least one)
User Experience:
- Delete button may be disabled on last persona
- Error message: "At least one persona required"
- Cannot complete deletion
Recovery Steps:
- This is by design - keep at least one persona
- Edit the remaining persona instead of deleting
- Modify it to desired default behavior
QA Note: Technical requirement, not recoverable error. User must maintain minimum one persona.
Error 5: Persona Configuration Too Complex
Cause: Multiple personas with conflicting settings or unclear order
User Experience:
- Responses seem random or inconsistent
- Unclear which persona is responding
- Conversation doesn't flow logically
Recovery Steps:
- Simplify persona configuration
- Reduce number of active personas
- Clarify persona instructions to avoid conflicts
- Test with fewer personas first
- Gradually add complexity
QA Note: Not technical error - user-created complexity. System works but results unsatisfactory.
Version History
| Date | Version | Author | Changes |
|---|---|---|---|
| 2025-10-04 | 1.1 | Iternal Technologies | Initial comprehensive documentation |
Notes
Important Considerations:
- Personas are chat-specific; each chat has its own persona configuration
- Template chats may include pre-configured personas
- Persona configuration is saved automatically with the chat
- Complex persona setups may slow response generation slightly
- Personas with dataset queries add processing time per message
Persona Context Types Explained:
- None: Standard AI responses based only on instructions and conversation history
- Static Text: Always includes specified text (e.g., company policies, reference info)
- Dataset Query: Automatically searches your dataset and includes relevant results
Context Placement Explained:
- Insert: Adds context as separate system message (AI sees it as distinct from user input)
- Prepend: Adds context before user's message (AI sees it as part of user's question)
- Append: Adds context after user's message (AI sees it as additional information)
Best Practices:
- Start with single default persona, add more only when needed
- Give personas clear, specific instructions
- Test personas individually before combining multiple
- Use descriptive names that indicate persona purpose
- Keep configurations simple unless complexity is required
- Document persona purposes for future reference
Common User Questions:
- "How many personas can I have?" - Technically unlimited, but practical limit around 3-5 for usability
- "Do all personas respond to every message?" - Depends on configuration; typically one at a time unless manually triggered
- "Can I use the same persona in different chats?" - Must configure separately per chat, or use templates
- "What's the difference between personas and templates?" - Templates create new chats with preset personas; personas are configured within a chat
- "Will personas remember previous messages?" - Yes, personas have access to full conversation history within the chat
Trigger
What initiates this flow:
- User manually initiates
Specific trigger: User wants to customize how the AI responds by creating or modifying personas, typically because:
- They want specialized AI behaviors for different purposes
- They need to simulate multiple perspectives or roles (e.g., legal team review)
- They want one persona for dataset queries and another for creative thinking
- They're using a template that includes multiple personas and want to adjust them
- They want to create a custom workflow or multi-agent system
User Intent Analysis
Primary Intent
Configure AI behavior by creating and managing multiple personas with different settings, contexts, and purposes to enable sophisticated, multi-faceted conversations.
Secondary Intents
- Simulate team discussions with different viewpoints
- Separate dataset querying from creative responses
- Create specialized AI assistants for specific tasks
- Build reusable conversation workflows
- Experiment with different AI behaviors
Subintents
- Understand persona configuration options
- Test persona behaviors before full use
- Organize personas for clear conversation flow
- Balance complexity with usability