Ideas on AI, organized as entries.

This is the readable archive: notes, research, framing, and working ideas about the AI stack from the physical layer up.

AI stack ground map

A first public version of the system map: power, chips, data, models, software, testing, and business context.

AI is not only software. Compute depends on electricity, cooling, sites, interconnect, and physical buildout timelines.

A model can only reason over what has been structured, linked, governed, retrieved, and trusted enough to use.

A practical guide for turning loose requests into context, constraints, examples, output formats, and verification steps.

A recurring note format for tracking what changed, why it matters, and which layer of the stack it touches.