Insights
This section collects short-form media: brief videos, observations, and exploratory ideas. These entries may not directly align to a specific system component, but they capture practical advice and emerging concepts relevant to Quarterly Systems.
Recent
▶️ From a Vague Prompt to an RPG… Sort Of
Date: 2025-10-21
Join us as we dive into the Torc × Vibes DIY Hackathon and show how to build an RPG using Vibes. In this walkthrough, Mike and Marcus demonstrate the Vibes workflow → from your very first prompt to publishing and remixing your game.
Along the way, we cover:
- 🔑 What makes Vibes unique as a full-stack AI app builder
- ✍️ The difference between high-fidelity and low-fidelity prompts
- 🐛 How to handle errors (and why bug fixing can be fun)
- 🌐 How to publish your app and use remixing as “version control”
- 🎭 Generating characters and demo data to bring your RPG to life
By the end, you’ll see how a single sentence prompt can evolve into a working-ish RPG scene… and why finding the sweet spot between too vague and too specific is a key to vibe coding.
▶️ I Stayed Up Late and Built My Own Stock Exchange
Date: 2025-10-21
I stayed up way too late and built a stock market for my own projects. Each ticker is something I’m working on (GG, $VIBE) and the prices move based on progress of tasks I set. Suddenly I’ve got a little economy running on my laptop.
Related: The KmikeyM Project - The world’s first publicly traded person
Now I can go sleep.
▶️ I Ruined Everything, But I Didn’t Give Up
Date: 2025-10-21
I tried building an app with AI… and it went off the rails. One bad prompt, and my boss literally quit the project on the spot.
This video is the story of how we were vibe coding together, how my design request blew everything up, and why I refused to stop. Even after Marcus bailed, I kept hacking, debugging, and pushing through until I had something that actually worked.
🔑 What you’ll see:
- How vibe coding can spiral from “magic” to “chaos”
- Why vague design ideas can break your entire app
- What happens when you stick with it after everyone else gives up
If you’ve ever broken an AI app with a dumb request, or just kept building after someone told you it was a waste of time, then we should hang out.
▶️ Claude + Obsidian Changed How I Think
Date: 2025-10-21
I’ve been documenting my life in notebooks forever and I love Obsidian. New notes for every day, plus backfilling from old notebooks has given me 900+ notes which is a span of 18 years of data.
When I connected Claude Code to my vault with proper architecture, something clicked:
- 45-minute daily planning sessions became 30 seconds
- Pattern recognition across decades, instant
- Crisis management highlighted in real-time
This isn’t about “AI-powered notes.” It’s about building an operating system for your life.
The three layers:
- Framework Layer (SKIF, PSP, TFRPP protocols)
- Data Layer (daily notes, call logs, health tracking)
- Intelligence Layer (Claude Code with session initialization)
What I cover:
- How Claude reads my Mission.md and Priority Stack Protocol at session start
- Live SKIF Strategic Brief generation (analyzes last 7 days in 30 sec)
- How Claude pointed out the physical impact of a housing crisis (and what I did about it)
Resources:
- 🔗 K5M Obsidian Template + Starter Files (see video comments)
▶️ Branch: Building Clubware, Not Software
Date: 2025-10-21
I built Branch: a tool to find vibe coders on GitHub. But I learned something building it… I wasn’t looking for a network map. I was looking for friends.
This is what vibe coding lets us do: build “clubware,” which is the small, weird software for ourselves and our friends. Apps that don’t need to scale to be successful.
If you’re building software like this, get in touch. Let’s hang out.
Links:
- 🔗 Branch repo
- 🔗 My GitHub
▶️ Don’t Let Claude Debug With More Code
A quick tip on working effectively with Claude Code.