HushMind
A mental health check-in app whose AI is trained to explicitly say 'I don't have enough to go on, can you tell me more' or 'this is beyond what I can safely help with' — calibrated to ask before it answers.
Adults who use AI for everyday mental health support and want honest limits instead of confident guesses
- Explicit 'I need more info' responses when the AI's confidence is below a set threshold
- Beyond-my-scope handoff that routes users to crisis lines, therapists, or emergency services when a topic warrants
- Session memory that shows what the AI does and doesn't know about you so check-ins build rather than reset
- Weekly reflection prompts that flag inconsistencies and ask clarifying questions before drawing conclusions
Mental health chatbots like Woebot and Wysa have scaled fast, but the Safe-Psych benchmark shows even strong models over-diagnose under incomplete information — users are increasingly wary of confident-but-wrong advice in a sensitive domain.
Chatbot-based mental health apps market valued $1.78–2.33B in 2024 and projected to reach $7.5–11.9B by 2033 at 15–20% CAGR — robust, growing demand.Chatbot-Based Mental Health Apps Market Report 2034 ↗Chatbot-Based Mental Health Apps Market (2025 - 2033) ↗
Crowded field — Woebot, Wysa, Replika, Youper and ~10+ named competitors already serve CBT/companionship use cases; the explicit 'ask-before-answering' calibration is a narrower niche but not unoccupied.Mental Health Chatbot Services Market Hits New High | Major Giants Woebot Health Wysa Youper Replika ↗AI-Based Chatbots for Mental Health: Comparative Study ↗
Established willingness to pay exists — Wysa premium ~$74.99/yr, coaching from $19.99, typical AI therapy subscriptions $15–80/mo — but the category is flooded with free tiers limiting ARPU.Wysa App Review 2025: Pros & Cons, Cost, & Who It's Right For ↗AI Therapy App Costs: Complete 2025 Pricing Guide ↗
Mental health is a durable secular need and rising regulatory/liability pressure (FDA scrutiny of clinical claims, Replika-style crises) structurally favors honest-limitation positioning, though compliance burden is high.AI for Mental Health 2026: Therapy Chatbots, Companions & Research ↗
Buildable on existing LLMs with prompting/safety wrappers and refusal calibration; no novel infra needed, but producing clinically defensible calibration data and navigating health-app regulations adds meaningful effort.What Are the Pricing Models for Mental Health AI Therapy Support? ↗