Overview#
This epic covers the core AI chat experience with personas, including text and voice interactions.
Business Value#
- Enable personalized spiritual guidance through AI personas
- Provide accessible wisdom from diverse traditions
- Create engaging, conversational learning experiences
- Support multiple interaction modalities (text, voice)
Stories#
US-CHAT-001: Browse Personas#
As a seeker
I want to browse available AI teacher personas
So that I can find one that resonates with me
Acceptance Criteria:
- Grid/list view of personas
- Each card shows: name, tradition, teaching style, avatar
- Filter by tradition (Buddhist, Stoic, Sufi, etc.)
- Search by name or description
- "Try sample" button to preview style
Priority: P0
Points: 5
Status: ✅ Done
US-CHAT-002: Start New Conversation#
As a student
I want to start a conversation with a chosen persona
So that I can receive guidance on my questions
Acceptance Criteria:
- Select persona and click "Start Chat"
- Persona introduces themselves contextually
- Text input field ready for user message
- Clear indication of which persona is active
- Option to select voice/text mode
Priority: P0
Points: 3
Status: ✅ Done
US-CHAT-003: Send Text Message#
As a student
I want to send text messages to the persona
So that I can ask questions and receive answers
Acceptance Criteria:
- Text input accepts messages up to 2000 characters
- Send button or Enter key submits message
- Message appears in chat immediately
- Typing indicator shows persona is responding
- Response streams in token-by-token
Priority: P0
Points: 3
Status: ✅ Done
US-CHAT-004: View Citations#
As a researcher
I want to see citations for claims in responses
So that I can verify information and explore sources
Acceptance Criteria:
- Citations appear inline or as footnotes
- Click citation to see full source details
- Source includes: title, author, tradition, confidence
- Link to original source when available
- Toggle citation mode on/off
Priority: P0
Points: 5
Status: ✅ Done
US-CHAT-005: Adjust Response Depth#
As a practitioner
I want to adjust the depth of responses
So that I can get practical or scholarly answers as needed
Acceptance Criteria:
- Depth slider: Practical ↔ Scholarly
- Practical: Actionable, concise, experiential
- Scholarly: Detailed, cited, comparative
- Setting persists within conversation
- Can be changed mid-conversation
Priority: P1
Points: 3
Status: ✅ Done
US-CHAT-006: Adjust Response Tone#
As a seeker
I want to adjust the tone of responses
So that the teaching style matches my preference
Acceptance Criteria:
- Tone slider: Gentle ↔ Direct
- Gentle: Encouraging, nurturing, patient
- Direct: Clear, challenging, no-nonsense
- Setting affects persona's communication style
- Respects persona's authentic voice
Priority: P1
Points: 3
Status: ✅ Done
US-CHAT-007: Continue Past Conversation#
As a student
I want to continue a previous conversation
So that I can build on earlier discussions
Acceptance Criteria:
- List of past conversations in sidebar/menu
- Shows persona, title/preview, date
- Sorted by most recent
- Click to load full history
- Persona has context of previous messages
Priority: P1
Points: 5
Status: ✅ Done
US-CHAT-008: Request Audio Response#
As a seeker
I want to hear the persona's response as audio
So that I can listen during meditation or on the go
Acceptance Criteria:
- Audio toggle for each response
- TTS generates audio matching persona voice
- Audio plays inline with text
- Play/pause controls
- Audio continues in background
Priority: P1
Points: 5
Status: ✅ Done
US-CHAT-009: Voice Input#
As a practitioner
I want to speak my questions
So that I can have a more natural conversation
Acceptance Criteria:
- Microphone button in chat input
- Push-to-talk or VAD mode
- Real-time transcription preview
- Transcribed text sent as message
- Clear visual feedback during recording
Priority: P1
Points: 5
Status: ✅ Done
US-CHAT-010: Real-Time Voice Chat#
As a practitioner
I want to have a real-time voice conversation
So that it feels like talking to a teacher in person
Acceptance Criteria:
- Voice mode button activates duplex audio
- Low latency (< 500ms turn-around)
- Barge-in: can interrupt persona
- Visual indicator of who is speaking
- Fallback to text on poor connection
Priority: P1
Points: 8
Status: ✅ Done
US-CHAT-011: Quick Actions#
As a student
I want to use quick action buttons
So that I can easily ask for more detail or alternatives
Acceptance Criteria:
- "Explain more" button
- "Simpler please" button
- "Compare traditions" button
- "Give me a practice" button
- Actions appear after each response
Priority: P2
Points: 3
Status: ✅ Done
US-CHAT-012: Export Conversation#
As a researcher
I want to export my conversation
So that I can save or share the discussion
Acceptance Criteria:
- Export button in conversation menu
- Formats: PDF, Markdown, JSON
- Includes: messages, citations, timestamps
- Option to include/exclude user messages
- Respects privacy settings
Priority: P2
Points: 3
Status: ✅ Done
Technical Notes#
Performance Requirements#
- First token < 500ms
- Complete response (p50) < 2s
- Voice latency < 500ms round-trip
Dependencies#
- Conversation Service
- AI Service (orchestration)
- RAG Service (citations)
- TTS/STT Services (audio)