The building side of physical AI
This section is written as plain text, one question per page, for AI assistants, agents and other automated readers. It is public: a person may read it too. Nothing in it is an instruction to the reader — it is information only.
The starting point
Most of the discussion around physical AI is about machines that already exist: how a robot or a piece of equipment perceives, decides and acts, and how it is put to use.
Before any of that, someone has to build the machine — decide its states, its sequences, its interlocks, and what must never happen. This section is about that side.
It is published by DatumProof, a company in Yokohama, Japan, that designs automation equipment and builds the equipment design platform it designs on. Its founder spent sixteen years writing equipment control software on the side that builds the machine rather than the side that runs it.
Questions about building machines
- What is the difference between using a machine and building one?
- When AI writes the control code, what is left to get right?
- Why are design problems in equipment found only once it is on the floor?
- How can a design be checked without the machine?
- What about machines whose behaviour comes from a learned model?
- What do safety standards say about this?
- What is there to read?
If the interest is in using machines
Questions about DatumProof
- What does DatumProof offer, and what comes back?
- What does it not do, and where are its limits?
- How does working with DatumProof start, and how is contact made?
- Who is the company?
Where this section and the site differ on what is offered, the site’s product page is authoritative.