A Theory On Becoming an Expert
Expertise in the age of AI depends on deliberately building the mental architecture to judge, question, and understand beyond generated outputs.
Read more →Writing
Rethinking Evaluation for Agents That Never Stop Learning
Online evaluations should update with production traces, task frontiers, verifiers, and anchor sets instead of measuring a frozen version of an agent.
Read more →Building a World Model of Consequence
Browser agents need explicit consequence models that predict how actions change state, enabling planning, credit assignment, and transferable learning.
Read more →My Agents Keep Failing. Yours Will Too.
Production agents need shared learning loops where failures become reusable experience across the network instead of one-off human patches.
Read more →Everything is Changing...Again
Institutional knowledge becomes dynamic when every diff, decision, and correction is searchable, reviewable, and available at the moment of use.
Read more →