Every week a new machine or model arrives. I go looking for the part no one owns yet.

Most writing about frontier tech stops at what’s new. I start where it gets interesting: once a trend takes hold, someone tries to fence the valuable part with a patent — and usually ends up holding far less than the headline suggests. The space between what a company wanted and what it actually holds is where the next invention still fits. That gap is what I write about.

What I write about

New models, chips, robots, tools — the arrivals are easy to find. Harder to find is where the ground is still open: which idea is genuinely locked up, which only looks locked, and which is anyone’s to build. I read the trend, then I go to the patents behind it — the one place a company has to state, on the record, exactly what it claims to own — and I mark the acre no one has fenced yet.

How I work

A patent is published twice: once as an application (what the company asked for) and once as a grant (what the examiner allowed). The difference between the two is what the company actually owns — and almost no one reads it. I put the two side by side and show the diff. Everything here says where it came from and when I checked it.

Why I write

I don’t have the answers to most of this. Writing is how I go and find them — reading the primary documents and showing the work. And every post ends with a list of what I could not confirm. That list is usually the most useful part.

A bit about me

I hold a Ph.D. in computer engineering and have spent twenty years — still counting — at a research institute, working on one problem from every side: physical AI, and the on-device, edge computing that lets it run in the real world. Reviewing inventions is part of my work, and I read patents closely — the way some people read box scores. These are my own views.

This is commentary on public documents, not legal advice, and it makes no determination of infringement.