Solving everyday problems with vibe coding #1

🚩 Problem

Too many meal choices causing daily decision fatigue.

💡 Solution

Helps users reduce decision fatigue by quickly suggesting a meal option through a simple web-based tool.

⛏️ Tools used

  • PRD: Google AI Studio
  • AI Agent: Lovable
  • API: OpenAI, Mapbox

Try Dishcision 👉 https://dishcision.lovable.app/

🤔 How I built Dishcision with Vibe Coding

1. Writing the PRD with Google AI Studio

I used Google AI Studio to write a clear PRD. Instead of just saying “make this,” I explained what I wanted to build, what problem it solves, who the users are, and the context behind it.

I asked specific questions to guide the planning:

  • What are the strengths and weaknesses of this idea?
  • What should I consider before building?
  • Will this be useful for the target audience?
  • How could this tool grow if developed in stages?

One useful trick: I asked the AI to put the PRD in a code block so it was easier to copy. This is especially helpful for non-developers.

2. Building with Lovable

  • Pasted the code into Lovable to build the UI.
  • The initial design wasn’t ideal, but improved it through multiple chat-based iterations.
  • Lovable integrates with 21st.dev, making it easy to update the visuals by reusing shared components.
  • Connected the tool to a basic map API to make it feel functional.
  • Future potential: Adding Google Cloud Places API or other location-based features could make it more context-aware and user-specific.

3. Integrated OpenAI API

I also tested a basic chatbot using OpenAI. It added a layer of interaction and showed how even small tools can offer personalized help.

🌱 Takeaways

What worked

  • Writing a PRD with AI helped me clarify the problem, not just the idea.
  • Seeing quick results kept me motivated to keep building.
  • The integration between Lovable and 21st.dev made it easy to apply and adjust code directly.

Challenges I faced

  • Simple text prompts weren't enough to accurately control or fine-tune the UI layout.
  • Repeated design tweaks through chat sometimes led to losing direction.

What could be improved

  • High-quality UI is still hard to achieve. More customization options would help.
  • Some AI-generated code didn't run as expected or didn’t apply cleanly in Lovable.