- Deploy backend to Railway (or Render for free tier)
- Deploy frontend to Vercel
- Update Google OAuth redirect URIs for production URLs
- Update backend CORS origins to allow production frontend URL
- RAG embedding support for Gemini/Anthropic (currently OpenAI-only for embeddings)
- Mobile responsive layout
- Email search/filter in inbox
- Pagination (load more emails on scroll)
- Draft history — view previously generated drafts
- Feedback analytics — dashboard showing rating trends over time
- Multiple Gmail account support
- Attachment handling (view and include in context)
- Custom system prompts — let users customize the LLM's reply style/tone
- Auto-refresh inbox on interval
- Better error handling and retry logic for Gmail API token expiry
- Loading states for thread view
- Keyboard shortcuts (j/k to navigate emails, r to reply)
- Email compose (not just reply)
- Unread/read status tracking