Ember.

Ember is an AI-driven burnout-prevention app for European corporate knowledge workers (25–40) in high-pressure roles — tech, finance, consulting, healthcare — who feel "always on" but haven't sought therapy. Instead of waiting for stress to hit, Ember detects patterns and intervenes before burnout through personalized AI journaling, pattern recognition, and contextual micro-prompts.

The product wedge is prevention, not reaction. Competitors like Bear Room help after stress hits; Ember's flywheel starts from day one with a 2-minute onboarding that produces an immediate Energy Level baseline and Buddy reflection. Each day the user does a 5-emoji mood check-in, gets a context-aware journal micro-prompt (one LLM call, Haiku-level), and receives a tiered AI reflection that grows in depth as the data accumulates — solving the 52% wellness-app churn that happens in the first week when generic reflections fail the "wow moment" test.

Powered by a two-model Claude strategy: Haiku 3.5 for fast, cheap journal prompt generation (one call per check-in) and Sonnet 3.7 for reflections (better reasoning, warmer tone, stronger grounding in user history). Claude was chosen specifically over GPT-4o and Gemini for Constitutional AI safety in a regulated mental-health context, alignment with GDPR / EU AI Act considerations (mental health AI is classified high-risk in the EU), and tone fit.

Targeting the AI mental health solutions market: $1.82B (2025) → $2.42B (2026) at ~33% YoY growth, projected to reach $6.7B–$25B by 2030–2032 at 30–40% CAGR. B2C iPhone-first, European-focused, designed for the corporate knowledge-worker context where burnout is an occupational phenomenon — and where regulatory readiness is a structural moat against US-only competitors.