Singapore One-Person Company Accounting: I Decided to Just Build It
Instead of a $60/month accounting service, I built a CLI-based back office and handed it to an AI agent. Took three days.
13 posts
Instead of a $60/month accounting service, I built a CLI-based back office and handed it to an AI agent. Took three days.
A persona isn't a tone of voice, it's a judgment policy. A record of an experiment extracting my own thinking patterns into a Persona Skill.
Dashboards are in fashion, but separating decisions into ADRs and execution into a Roadmap fit better. A story about the structure I chose for repeatable quality, against a backdrop of SKILL-first design and AI loops.
This one isn't about writing technique. It's a record of taking something I actually did — realigning my persona — and pinning it down as an automatable skill.
A conversation with Yum in Singapore. Will 'good value-for-money cities' survive the AI era? Where will humanity's 10,000-person red team gather?
That AI isn't 100% doesn't mean humans are needed. Coding is no longer transportation — it's a marathon.
I ran a religious argument with ChatGPT all the way down and the structural difference between science and theology came into view. Also, the time I nearly built an AI cult leader.
Autonomous open source where every agent is a PR author, a validator, and a voter. I'm still working out how to think about merge power and hostile takeover in that structure.
An autonomous system where every agent is a PR author, a validator, and a voter. How should meta-governance and the balance of power be designed in that structure?
Between coder and architect in the AI era. The MoltBot creator's line about shipping code he hasn't read made me slap the table.
These days 90% of my work runs through Claude Code. Figured a blog would fall out of it just as fast, so I built one.
Pulling my persona out of ChatGPT conversations and getting AI to write in my style. A thinking-style persona plus purpose-specific personas.
Why semiconductor design needs AI agents. Working through the structure that lets LLMs perform optimization in the discontinuous space of code.