System design

The monorepo was always a bet on shared understanding. In the AI era, that bet pays exponentially — if you evolve what lives in the repo alongside the code.

Most teams dump everything into a vector database and call it knowledge management. The ones that win curate instead — signal over noise.

AI executes well, originates poorly. Personal operating systems are how structured expertise compounds into something a machine can act on, and scales to teams.

Each paradigm shift in software has been architectural. AI agents are the next — but only if we design for them intentionally, not as browser overlays.

Explore the emerging divide between cloud and on-device AI, where Google leads large-scale intelligence and Apple builds a private, local AI revolution.

AI’s future lies in vertical agentic systems, multimodel pipelines that blend context, automation, and human expertise to amplify real business impact.