I write journals, I give talks, I sit in meetings, and all of it gets transcribed somewhere. Together that’s plenty of evidence about how I actually spend my attention. I want agents that read it and tell me something I’m not seeing.
The design I landed on is one agent per value. I have a set of values I’ve been working with for years: legacy and making a difference as the core, with fairness, gratitude, humor, and others around them. Each agent reads my journal entries, talk transcripts, and meeting notes, looking for one value. Then it tells me how I’m living into it, where I’m diverging from it, and where I had an opportunity to practice it that I missed.
Then a second agent synthesizes across all of them. It connects the values to the projects I’m in, the people I’m working with, and the ideas I’m developing. It might point out that the same person keeps showing up when I talk about patience, or that a project I’ve been avoiding is the one carrying my legacy value.
When it should arrive
I don’t need this every day. I need it when I’m already thinking.
The useful moment is my daily or weekly review, when I’m pondering what the week was about. That’s when a short note that says “here’s what your week showed about how you lived your values” would land. Weekly, monthly, quarterly: whatever cadence I ask for, delivered when I’m already reflecting.
The goal isn’t a dashboard. It’s reaching connections faster than I would on my own. I’m not going to reread hundreds of journal entries looking for a pattern. An agent will.
At the end of the year
Next December, when I start preparing my annual review, I want to point these agents at those reflections and ask what changed. Maybe Q1 was about making a difference and Q2 was about something else entirely. Not what I said I was focused on, but what the activity shows.
Patterns like that are hard to see when you’re inside them and easy to see in a year’s worth of writing.
Goals belong in this too
The last piece is goals. I set goals every year, and I’d like to see which values each one actually serves, mine and Improving’s. A goal that serves no value I care about is worth questioning. A value with no goal attached might be the one I’m neglecting.
What this really is
This is a thinking tool. It reads what I’ve already produced, finds the threads I didn’t notice, and hands them back to me when I’m ready to look. The journaling is the raw material, and the agents are the second reader.
I’ve started smaller with an agent that processes each journal entry as it arrives: splitting it by topic, pulling out to-dos and blog ideas, and linking people and threads to their notes. The values agents are the next layer up. I’ll build the per-entry one first and let it prove itself before adding the bigger picture.
Leave a Reply