Regeretably

• 

  Demo checkout page (simulated door dash)

• 

  Summary of all your recent purchases and identified as potentially regrettable

• 

  roast message that appears in the corner

• 

  modal asking if you still want to proceed

Inspiration

This project was inspired by a mix of financial regret, internet culture, and the ever-growing interest in using AI to improve everyday decision-making. I realized that many of my impulse purchases — late-night takeout, overpriced gear, or trendy gadgets — often left me with a sense of “why did I buy that?” I thought, what if I had a brutally honest voice in my head to stop me before making bad money moves?

Naturally, that voice had to be Andrew Tate — someone known for confidently (and controversially) roasting poor decisions. The idea was to combine personal finance awareness with humor and personality, turning a dry budgeting app into something that you will actually listen to.

What I Learned

I learned more about voice generating APIs and how to create custom trained voice models that sound realistic, engaging, and timely enough to influence decision-making in real-time.

How I Built It

I built the project using:     •    Next.js for the front-end framework     •    Eleven Labs for realistic voice cloning and audio generation     •    OpenAI API to analyze user behavior and evaluate spending decisions     •    Node.js for server-side processing and backend integration

Challenges Faced

Some of the main challenges included:     •    Training a quality, custom voice agent that sounded natural and roasted users without sounding too repetitive or artificial     •    Syncing AI logic with voice timing, ensuring the roast happened before the user committed to a purchase     •    Balancing the tone so it was entertaining but still helpful, without crossing into harsh or offensive territory

Built With

• knotapi
• next.js
• typescript