A collection of my thoughts and stuff.
Recently I’ve been doing most of my personal development work on VMs from https://exe.dev. More on why in a future post, but in summary; these long lived VMs offer a safe way to run agentic coding tools like Claude Code in --dangerously-skip-permissions mode, which makes the models really shine.
My first thread in Amp is from 7 months ago, July 2025. Back then, Claude Sonnet 4 was the model used by Amp’s smart mode. Today that is Claude Opus 4.6. Much has changed in that time - the models have become smarter, and I’ve leaned into AI assisted software development more.
For the past few months I’ve been using AI coding agents heavily. So far I’ve used: AmpCode - I started out mostly with their VSCode extension & then moved exclusively to the CLI Claude Code - CLI Codex - CLI OpenCode - CLI Codex - App
This will be a very short post to put something on the internet that I couldn’t easily find with Google.
I’ve completed a freelance project I was working on for a few months, and have started saying no to new opportunities. It’s time to work on one of my own ideas again. This is part of my plan to start failing more.
As part of a contracting project, I’ve been building an analytics dashboard for a feedback collection SaaS. The app is built in Ruby on Rails and given all the nice things I’ve heard about Kamal; I decided to use it for deploying the app.
I have failed, and that is exactly what I had hoped for a few months ago in this blog post.
If you’re just looking for implementation instructions, skip my ramblings and go straight to the code here.
Links:
I was recently listening through the How to succeed at failing series on the Freakonomics podcast and started to think about how often I had failed in the past few years. The first answer was - not too much. I couldn’t think of too many instances of where I had “failed”.