
I have an “agent with character” that sits in the root of my Obsidian vault and helps me with a ton of stuff.
Ziggy Ragdoll knows I somehow got hoodwinked into being Head Coach of my son’s t-ball team. I signed up Assistant Coach. They told me they just needed a Head placeholder in the TeamSnap system. Next thing I know, I’m doing batting orders.
This Side Quest is part of a recurring theme where AI handles my behind-the-scenes. I worked with the Vault agent to create a Skill that generates batting and fielding orders for batches of 3 games that are designed to give all kids equal time at all fielding positions (and batting orders), making sure everyone gets one of the treasured ones each game (Pitcher and 1B, obviously). Then gave Ziggy access to the Skill, and she takes it from there.
In practice, the flow is: I communicate with Ziggy over Signal, Ziggy edits a Google Sheet in near-real-time when we have no-shows (which are always last minute for some reason), and my fellow coaches and parents can view along.

The outcome is that when I’m on field, I get to go full Ted Lasso and focus on the kids.
So What’s the ROI?
Probably saves 15 minutes a game doing a lineup and responding to last minute absence. Trade that off against me jamming with Claude for a bit (say 2 hours) to set up the agent, learn about Google Sheets connectors etc. So the up-front investment evened out roughly mid-season.
But is the AI ROI just Rob’s time here? Or the quality of outcome, which I’m not sure I can quantify?
Either someone else would have done all the setup, or we would have had a less equitable (and notionally less fun) outcome.
Full Ted Lasso. That’s the ROI.