DecideNow

The Problem

Every day, millions of people sit paralyzed in front of their to-do list. Not because they're lazy — but because when everything feels urgent, the brain shuts down and picks nothing.

Existing productivity tools make this worse. They assume you're already organized, already focused, already know what to do. They ignore the most important variable of all: how you actually feel right now.

What I Built

DecideNow is an AI-powered decision tool that solves the exact moment of paralysis — when you have 3-4 tasks and genuinely cannot pick one.

Here's the full flow:

1. Mood Check — Tell the AI how you're feeling. Exhausted? Energized? Stressed? This changes everything about the decision.
2. Task Input — Dump in 2-6 tasks you're stuck between. No organizing required.
3. 3 Clarifying Questions — AI asks 3 short questions tailored to your mood and tasks.
4. The Decision — AI picks one task with clear reasoning based on your energy level. Not a suggestion. A decision.
5. Message from Future You — Unlock a personal message from your future self 2 hours from now, after completing the task. Specific, emotional, and real.
6. Accountability Check-in — Come back when done. Tell the AI how it went. It responds like a real person.
7. Brutal Honesty Mode — Toggle this on if you want the AI to skip encouragement and tell you the hard truth.
8. Procrastination Detector — If you keep restarting without completing tasks, DecideNow notices and calls you out.

What Inspired Me

I kept building to-do lists and never finishing them. The problem wasn't the list — it was that I never knew which thing to actually start. I wanted a tool that made the decision for me, not just organized my chaos.

How I Built It

• React + Vite for the frontend
• Groq API (Llama 3.3 70B) for fast, free AI inference
• Custom CSS with dark theme
• Vercel for deployment

The core challenge was prompt engineering — getting the AI to consistently return clean JSON across all prompts while making mood actually change the output meaningfully, not just cosmetically.

What I Learned

The most impactful feature wasn't the decision engine — it was the "message from future you." It turns a utility tool into something that feels genuinely human. Emotional design matters as much as functional design.

Challenges

• Getting Groq to return consistent JSON without markdown formatting required careful prompt engineering
• Making the mood layer actually influence decisions meaningfully took significant iteration
• Balancing brutal honesty mode to be direct without being demotivating

What's Next

• Browser notifications for check-in reminders
• History of past decisions and completion rates
• Team mode — make decisions together

Built With

• css
• groq
• llama-3.3-70b
• react
• vite