Bernie Goldbach

My dirty notes and my lake of knowledge

· Bernie Goldbach
Bernie in his comfy winter gear in 2019

I run a set of scheduled tasks with Claude every weekday morning and they generate important Notes to Self worth reviewing. I use Anthropic heavily before 11AM in Ireland for these tasks. After that time, it feels like I burn more AI tokens to get the same amount of work done. Then I get my rights of heavy usage back just after 6PM.

My Dirty Notes

I read the outputs generated by Claude just like I would read the Dirty Notes that inspectors on USAF flight lines would generate during major inspections. If an inspector trusted your methods and processes, you could get a few scribbled back of the envelope snags to fix. Those dirty scribbles (Dirty Notes) would never appear in the final written reports.

Several Dirty Notes require me to change how I train people and they link directly into the IRDGAI and ManagerAI Playbooks that I teach. Another specifies concerns about some of the content I’m using online. I got an email from a Chinese Domain Registrar about the insideview.ie website I have run since the days of my Friday column with the Irish Examiner. A Chinese company is exploring the prospect of a global trademark with the phrase “insideview” (all one word).

Another Dirty Note politely advises action concerning the US TAKE IT DOWN Act and points out that 46 US states now have laws targeting AI-generated synthetic media. I have some AI-generated still images on the Iteration website that I want to change to photos I’ve taken.

My skills and my lake

Claude cited a phrase I use during workshops: If you’re doing the same thing twice with Claude, you should create a skill" because Claude knows I have a calendar item “update microblog” and that suggests I should make a Claude Skill that dumps snippets from Dirty Notes onto my insideview.ie blog. I don’t think I will do that. I want people to think it’s Bernie on the keyboard creating this content.

I’ve asked Claude what it knows about me and it confidently said that I have a the knowledge base with Obsidian that is “a lake” that evaporates without “rivers” (recurring ingestion pipelines) from my Fabric and my Inoreader. I need to write a Claude Skill that ports important content from both of those services into my Obsidian Vault. The live dashboard I use as a health check service of my Personal Knowledge Management system is only as good as the pipelines feeding it and right now the Fabric and Inoreader pipelines are silos.