Everyday

Inspiration

I wasn't managing my life well, and I wasn't happy with where I was. I wanted to improve — but improving anything starts with actually understanding where you are, and I didn't have that. If I wanted to build muscle, I needed real data: where am I starting from, what am I actually eating, is it working. As a guy trying to get his life together, money pressure is real too — without real budget tracking, that pressure only gets worse.

That's where the idea started: I needed data, not vague intentions. So I decided to build an app that tracks everything — and since data alone doesn't give you insight, connect it to AI through MCP so I could actually ask questions about my own life and get real answers back.

What it does

• It's a giant application containing around 40 rich, single-purpose features — including budget management, calorie tracking, investment tracking, and more.
• No server required after the initial setup.
• It has a working MCP server on an Edge Function that serves all this data to any AI for analysis.

How we built it

Everyday was built entirely through Codex, using GPT-5.6 Terra. The only thing i do was just prompting

Challenges we ran into

• Supply chain safety — I didn't want to accidentally install a typosquatted package.
• MCP server configuration ran into a chain of real 5xx errors before it actually worked.
• Documentation was harder than expected — even with AI writing it, I had to review everything myself, and there was a lot of it.
• Recording the demo video at 3am almost got me a knock from my neighbor.
• Every feature took multiple rounds of prompting to get to something actually production-ready, not just "looks done."

Accomplishments that we're proud of

• A working product — and the best part is I can actually use it, and will keep using it for myself.
• This is my first hackathon submission. Here we go.

What we learned

• In the age of AI, prompt engineering is a real skill — this whole application came from prompting alone, and it's genuinely useful, not a toy.
• Cost management is something I need to work on — a few days in, I'd already used close to $50 in AI credits. I need to get better with my prompting.

What's next for Everyday

• Right now I'm actually using it for my daily life, so honestly, I'm content just logging everything going forward.
• If this wins, I'd genuinely consider turning it into a real SaaS — Everyday!

Built With

• codex
• edge-function
• mcp
• openai
• react
• supabase
• typescript
• vite