💡 Core Idea Strength

Multi-persona AI with persistent memory per mode is a genuinely differentiated concept. The market mostly offers one-size-fits-all chatbots — mode-specific context is a real gap.

✅ Strengths

• StrongClear niche personas — each mode has a real user base
• StrongJEE Tutor is underserved & high-demand in India
• StrongMode-specific memory feels premium vs generic AI
• StrongCompanion mode creates retention & daily habit
• StrongMode switching = flexibility without losing context

⚠️ Critical Risks

• HardLong-term memory needs a backend (e.g. Supabase + vector store)
• WatchCompanion mode has ethical/emotional dependency risks
• WatchJEE accuracy must be very high — errors harm students
• HardVibe-coding alone won't handle memory persistence

🛠️ Build Priorities

• V1Mode switcher UI + basic persona prompts
• V1In-session memory (context window)
• V2Persistent memory via DB (per mode, per user)
• V2Auth + user profiles
• V3Cross-mode memory (optional, toggleable)

🔑 Key Tech Decisions

For vibe-coding, recommended stack:

• Frontend: Next.js or Bolt.new
• AI: Claude API or OpenAI
• Memory: Supabase (DB) + Mem0 or pgvector
• Auth: Clerk or Supabase Auth

🎯 Monetization Angle

Free tier with 3 modes → Paid unlocks all 5 + unlimited sessions. JEE Tutor alone could be a paid standalone product in India.