Physical AI: From Boundaries to Certainty

We turn hardware into software — and devices into actions. Start with boundaries. Not code. Just answers. So AI can act with certainty.

What is this?

Nemo & Anna is an action-centric approach to building Matter devices.

Instead of modeling devices by type, we define them as a set of Actions.

Each action is defined with:

• intent
• limits
• safety boundaries

How it works

Scan once — and AI receives the full execution model.

Learn more about the execution model and boundary system:

github.com

GitHub - Jang-woo-AnnaSoft/execution-boundaries: Design notes on execution boundaries and...

Design notes on execution boundaries and responsibility structures for AI systems interacting with the physical world.

From that definition, a real Matter device is generated on ESP32-C6 — ready to execute safely and predictably.

Why this matters

AI doesn’t guess what to do. It acts with defined intent and clear boundaries.

No code required

Just answer a few questions —

• 4 for a simple action
• up to 9 for full definition

Try it

Nemo & Anna User Guide

Project overview

https://anna.software

Explore the structure

GitHub - anna-soft/Nemo-Anna · GitHub