PS: Evaluating Agent Systems
Solving three problems at once.
The entire world is going agentic and within the bounds of my healthcare startup there are three questions:
How will humans working on their health employ agents to do the drudgery so they can focus on the high value learning and decision making tasks?
How can the junior support person tasks that arise when using PostHog to track customer experience be turned over to an agent system?
How can I take the junior support person duties found in operating Grafana, limited only to observation and escalation, creating an agentic system to shield the 24x7 support staff from alert fatigue.
I’ve been poking at the edges of this problem for a while, but on Monday I decided to solve it in an organized, systematic fashion. Come on in and see what I did …
