PhysMind CLI

The Industrial Physics Brain — Expert AI CLI for Manufacturing, EV R&D, Robotics & CNC
Built by Keplore AI · Silicon Valley

Install · physmind.ai · Rust Workspace

---

What is PhysMind?

PhysMind is the industrial physics brain built by Keplore AI. It achieves state-of-the-art performance on industrial AI tasks through expert agents — not bigger models.

The paradigm shift: instead of scaling foundation models on general data, PhysMind packages industrial expertise as executable skill packages — physical equations, calibration procedures, acceptance criteria — invoked at inference time by agents.

Industrial decision dependencies are not documents to retrieve — they are structured processes, physical equations, and acceptance criteria that must be executed.

Where PhysMind works:

• Manufacturing vision & defect detection
• EV battery & cell inspection
• Robotics integration & vision calibration
• CNC optimization & PLC integration

PhysMind CLI

PhysMind CLI brings the industrial AI brain to your terminal. Powered by Qwen via PhysMind's built-in proxy — no API key setup required.

• Works on macOS, Linux, and Windows
• First launch prompts for a kplr- access key (one time only, saved automatically)
• Streaming responses with thinking display
• Session history, tool use, bash execution (! prefix)

Install

macOS / Linux:

curl -fsSL https://physmind.ai/install.sh | sh

Windows (PowerShell):

irm https://physmind.ai/install.ps1 | iex

Quick Start

physmind

On first launch you'll be prompted for a PhysMind access key (kplr-...). Get one at physmind.ai/install. The key is saved to ~/.config/physmind/credentials — you won't need to enter it again.

Usage

physmind [OPTIONS] [PROMPT]

Options:
  --model <MODEL>       Model to use (default: qwen3.6-plus)
  --resume              Resume the latest session
  --output-format json  Machine-readable output
  --help                Show help
  --version             Show version

Run a bash command directly inside the REPL:

> !ls -la

Build from Source

git clone https://github.com/KeploreAI-Lab/physmind-cli-rust
cd physmind-cli-rust/rust
cargo build --release -p rusty-claude-cli
./target/release/physmind

Supported Platforms

Platform	Binary
macOS (Apple Silicon)	physmind-macos-arm64
macOS (Intel)	physmind-macos-x86_64
Linux x86_64	physmind-linux-x86_64
Linux arm64	physmind-linux-arm64
Windows x86_64	physmind-windows-x86_64.exe

Why Expert Agents, Not Bigger Models

Generic foundation models fail in industrial settings for four structural reasons:

1. Models learn patterns — industry requires principles. Physical laws and acceptance criteria can't be abstracted from text data.
2. Industrial data doesn't exist publicly. Calibration parameters, PLC quirks, factory protocols — none of this is in any training corpus.
3. Probabilistic output fails in safety-critical settings. A miscalibrated machine or failed inspection is not an acceptable error rate.
4. Per-customer fine-tuning is operationally impossible. PhysMind adds new verticals without retraining — by packaging expertise as executable skills.

About Keplore AI

Keplore AI builds PhysMind — the industrial physics brain. Based in Silicon Valley.

• Website: physmind.ai
• Support: support@keploreai.com
• Open source runtime: physicalflow-runtime