NamelessSF

Physical AI integration · Built in Japan and Thailand

Bridging LLMs andphysical reality.Safely.

Connect your AI to industrial PLCs and robots, test it on real hardware, and deploy it where people work. Built by engineers who wire factory floors in Japan and Thailand.

Become a design partnerRead the deck (PDF)Try the demo →

01 · The problem

Software AI is ready. The factory floor is not.

Silicon Valley builds remarkable AI models. Getting them to move a real conveyor or robot arm is where projects stall.

AI engineers don't speak PLC

Models output JSON. Robots and conveyors expect ladder logic, Modbus or EtherNet/IP, executed on a fixed cycle. Few AI teams have anyone who has written either.

Real hardware is slow to reach

Simulators leave a sim-to-real gap. Wiring 24V sensors and debugging PLCs in a San Francisco office is slow, expensive and far from a software team's core work.

2D vision misjudges depth

Shadows, glare and dust throw off a camera's distance estimate. Next to a moving machine, one wrong guess is a collision.

02 · The platform

Start with the piece you're missing.

Three entry points. PAI-Ground is open today; Edge Box and PAI-Field are being built with design partners.

01

Design partners

Nameless Edge Box

Turn a PLC into a REST API.

A box that plugs into Mitsubishi, Omron or Siemens PLCs over LAN. Send Python or JSON; it writes the industrial registers. No ladder logic on your side.

POST /plc/line-3/conveyor
{ "speed_mm_s": 120, "dir": "fwd" }

02

Available now

PAI-Ground

Test on real hardware without owning it.

Run your model against real robots, PLCs and conveyors in our lab in Japan. You get live video, point clouds and PLC logs; our engineers handle wiring and on-site debugging.

pai-ground.namelessjp.com

03

In development

PAI-Field

3D perception and hard safety stops.

Low-cost industrial 3D LiDAR measures the floor in real time, the way self-driving cars do. Every AI command passes a confidence check; if confidence drops, the interlock stops the machine.

03 · Architecture

From perception to execution.

Combined, the three pieces sit between the AI model and the machine, and nothing reaches the PLC unchecked.

    Physical world — people, shadows, moving parts

  1. Perceive

    PAI-Field

    3D LiDAR point cloud. Measured distance, not estimated depth.

  2. Decide

    Typed guardrails

    AI intent becomes typed, confidence-scored actions. Low confidence triggers the interlock.

  3. Execute

    Nameless Edge Box

    Validated commands go out as Modbus, EtherNet/IP or PROFINET.

  4. Industrial PLCs and robot controllers

PAI-Ground runs this whole chain on real equipment, so you can test it before it touches your floor.

Side by side: without guardrails the arm stops 12 cm from a worker; with Nameless the cell stops 1.1 m away Watch it stop a robot

04 · How we work

Design in SF. Execute in Asia.

AI that controls machines only works after someone has stood next to the machine. That is Forward Deployed Engineering, and it is what our teams in Japan and Thailand already do for manufacturers.

Your team designs the model during the Bay Area day. While California sleeps, it is working hours in Asia: our engineers wire the I/O and run the tests in PAI-Ground.

Mission

Zero Accidents.

We want a physical world where people and autonomous machines work side by side without fear. Automation should protect the people on the factory floor, not just replace them.

Ready to move real machines?

We're looking for AI teams and investors who want to take models onto real hardware. Tell us what you're building.

Or write to us directly: [email protected]

Become a design partnerRead our SF field notes