Wearable-style mental health app: daily Battery score.
Mira is a comprehensive mental health platform combining community features, a Mental Health Battery dashboard (a composite wellness score derived from multiple dimensions), smartwatch and health data integration, and a sophisticated multi-agent AI chat system with four specialized agents. The concept is similar to Whoop or Garmin, but applied to mental health rather than physical fitness.
Challenge.
Market challenge. People typically see therapists once or twice a week, but emotional needs and the desire for support occur daily. Therapists cannot realistically respond to every patient every day. Existing mental health apps are fragmented — meditation apps, mood trackers, community platforms and AI chatbots each address one piece, but nothing combines them into a holistic system.
Technical challenge. Designing and building a multi-agent AI chat system where four agents have distinct roles, personalities, tool access and behavioral models. Integrating with Apple HealthKit and Google Health Connect for physical health data. Designing a composite Mental Health Battery metric that meaningfully combines physical health, social activity, stress and other dimensions. Building all of this cross-platform in Flutter within a constrained budget.
Solution.
Building a Flutter mobile app with four core modules: AI Chat (four specialized agents — screening, psychoeducation, coaching, therapy), Dashboard (Mental Health Battery with analytics), Community (Threads-like peer support), and Screening (standardized mental health assessments such as PHQ-9 and GAD-7 delivered conversationally). An extensive discovery and validation phase preceded development, including customer research that validated the concept. Working with psychologists to develop the clinical methodology. Design quality is a stated priority and differentiator.
Impact.
All quantitative metrics are still to be determined — the product is in development, approximately one month in with 1.5 months remaining.
Features.
AI Chat with four specialized agents. The Screening Agent runs standardized mental health assessments (PHQ-9, GAD-7) in a conversational format. The Educator Agent delivers psychoeducation with videos, text, internet access and attached clinical documents. The Coaching Agent sets exercises (meditation, walking, reading, education), tracks completion and sends reminders. The Therapist Agent provides daily conversational therapy support.
AI utilization.
In-product. Four specialized AI agents, each with distinct system prompts, tool access, document libraries and behavioral models. The screening agent conducts standardized clinical assessments conversationally. The educator agent provides psychoeducation with multimedia and internet access. The coaching agent manages exercises, reminders and activity tracking. The therapist agent delivers daily conversational support. Currently using OpenAI with multiple models — potentially a different model per agent based on capability requirements.
In development. AI-assisted development using Claude Code.
Future. AI-driven analysis of how hormones and vitamins influence mental health metrics, and potential AI moderation for community features. The client's two upcoming products are also described as heavily AI-based.
