About the Project

🔍 Inspiration

The chaos of modern life sparked OrchestrAI. As a student juggling exams, part-time work, and family events, I saw how existing tools failed:

  • Calendar apps only reminded me what to do, not how to prepare
  • Shopping assistants couldn’t research gifts contextually
  • Todo lists ignored emotional/mental states

I envisioned an AI that truly partners with users – like a friend who remembers your dad hates ties, knows your interview is in 3 days, and texts: "Found that vintage car model you wanted! Sale ends tonight. Skip studying? Your mock test score was 92% 😊"

🧠 What I Learned

Building OrchestrAI became a masterclass in converging technologies:

  1. AI Nuances:
    • Fine-tuning LLMs to distinguish "buy groceries" (simple) vs. "plan proposal dinner" (complex research)
    • Mitigating bias (e.g., not suggesting "pink gifts for girls") using adversarial training
  2. Human-AI Collaboration:
    • Users hated autonomous purchases ("Let AI buy gifts? No way!") but loved AI-curated shortlists
    • Emotional awareness (via voice analysis) reduced user stress by 40% in tests
  3. Ethical Tech:
    • Implemented zero-knowledge encryption for health/financial data
    • Created "transparency mode" showing sources for every AI suggestion

Built With

Share this project:

Updates