For Thinking Machines, from 🤫 hussh
Intelligence that lives where people do.
You are building AI that extends human will and judgment, distributed and shaped by the people who use it. We are building the place that intelligence should live: on hardware a person owns, under their consent. Two halves of the same sentence.
Why Thinking Machines
Good science, in the open.
“For artificial intelligence to benefit from distributed knowledge, it must itself be distributed.”
Mira Murati founded Thinking Machines Lab in 2025. The team publishes its research in the open, ships the code beside the paper, and made fine-tuning something a small team can do with Tinker.
We read that line and recognized our own roadmap. Distributed intelligence needs distributed machines, and the most personal machines are the ones people own. 🤫 Private Agent One, Puppy One, Tag One and Factory One are built to be that home: private, sovereign, and always on.
Read on
Three short pages, written for you.
Their work
Tinker, Connectionism, the open-source code, and the through-line of the science, from Thinking Machines' own writing.
Together
Their published direction, theme by theme, beside what 🤫 builds and what it unlocks for both of us.
How we start
A provider-neutral bench and a teach-your-own-agent prototype first, then a pilot, then building and selling together.
The born and the made, on the same road
Your science of teaching a model, meeting the place that model should live. We would love to build it with you.