Kris Kang
I write about business, economics, AI, math, science, and philosophy. When I have time, I write here to reflect and understand.
Writing
Filter
- 2026-10-22the 1-person companyFrom the East India Company to Anthropic's agents: why firms built hierarchies, and why the agent-run firm may be the one that is almost all talk.
- 2026-10-07decision models are not newTypeSafe's Jev returns typed, calibrated decisions instead of text. Useful, cheaper, and not entirely new.
- 2026-09-18capital revisitedA shorter reading of Dwarkesh and Trammell on capital in the 22nd century, working through Piketty's model to the conditions under which AI could run capital's income share towards one.
- 2026-09-18iconographySoftware architecture can encode a division of labour into the structure of a system, until nobody carries the full expression of intent.
- 2026-09-09it takes navier-stokes to move the worldAn attempt to explain the Navier-Stokes excitement to myself, with AI assistance.
- 2026-09-03measuring real-world ai progressTwo asks: report benchmark performance against test-time compute, and publish harness ablations.
- 2026-08-01the missing ai infrastructureThe intelligence is already to hand. What remains to be built is the scaffolding of identity, boundaries, evidence and governance.
- 2026-03-19the triple debtCompanies adopting AI coding tools are accumulating three debts at once: financial, technical, and organizational.
