Text & voice conversation
Conversational reflection through supported text or voice modes.
Mjuzi AI Companion
Mente Mirror AI Companion - LIVA
LIVA is a conversational AI designed to listen before it guides—helping you reflect, notice possible patterns and practise manageable wellbeing actions without pretending to be a therapist.
How a conversation is designed
Mjuzi is designed as a guided conversation, not a rapid-fire assessment. It follows your concern, asks one primary question at a time and keeps deeper exploration permission-led.
- 01
Listen
Start with your words and your pace.
- 02
Reflect
Mirror what was shared without adding hidden meaning.
- 03
Explore
Ask one relevant question and invite correction.
- 04
Practise
Offer one approved low-risk tool when useful and wanted.
- 05
Integrate
Check what helped, note what remains, then close or connect onward.
Safety and boundaries come first. Crisis, safeguarding, medical urgency, requests for human support and prohibited clinical requests override the ordinary conversational flow.
Everyday capabilities
Useful support without pretending to do everything.
The experience groups common needs into a smaller set of clear capabilities.
Emotional reflection
Supportive acknowledgement, clarification and gentle pattern exploration.
Check-ins & continuity
Short check-ins and approved prior context can help a returning conversation start with less repetition.
Low-risk wellbeing tools
Approved options such as breathing, grounding, reflection or journaling where eligibility and consent allow.
Personalised next steps
Small suggestions informed by the current conversation and permitted wellness context.
Human-care pathway
Clear routing toward psychologists, psychiatrists or other appropriate support when AI should not decide alone.
AI + human, clearly separated
LIVA supports the conversation. Qualified humans retain clinical judgement.
The distinction is deliberate: immediacy and reflection can be supported by AI; diagnosis and consequential clinical decisions stay with qualified professionals.
LIVA can support
- Listening and reflective conversation
- One-question-at-a-time exploration
- Approved low-risk wellbeing tools
- Check-ins and approved continuity
- Drafting bounded summaries where configured
- Guiding you toward human care
Qualified professionals handle
- Diagnosis and clinical conclusions
- Medication and treatment decisions
- Complex or high-risk clinical care
- Restricted personalised interventions
- Safeguarding and critical professional judgement
- Approval of client-facing reports where required
Continuity without a hidden profile
What LIVA may remember—and what it should not.
Returning sessions are designed to use the smallest relevant amount of approved context rather than replaying your history or turning AI guesses into facts.
Useful continuity
Approved summaries, goals, preferences and tools you have tried may be used when relevant and permitted.
No durable hidden psychology
AI hypotheses, inferred motives, personality labels and theory-derived conclusions are not meant to become ordinary long-term memory.
Your current words matter most
If something has changed, your current account should take precedence over stale context. Disputed information should not be treated as settled fact.
Control follows consent
Eligible information can be corrected or excluded, and revoking an optional consent should stop future use within that scope.
After a session
Facts, professional interpretation and your summary stay separate.
Where the post-session report workflow is enabled, Mente Mirror is designed to keep three information layers distinct so an AI-generated interpretation does not silently become a client fact.
Factual session history
What you actually shared: chronology, context, strengths, goals and selected exact wording, with missing information left as missing.
Grounded first · reviewed before approved longitudinal useClinician-facing formulation
Tentative psychological hypotheses remain separate from facts, include uncertainty and are restricted to authorised clinical review.
AI draft is tentative · formal diagnosis stays human-controlledClient-safe post-session summary
A concise summary can include important themes, strengths, approved tools, reflective questions and next steps.
Human approval before release where this workflow appliesPrivate clinical material stays private. Clinician-only notes, hidden hypotheses and detailed safety records are not intended to appear in an ordinary client-facing report or conversational memory.