Anthropic has unveiled a research framework designed to allow artificial intelligence agents to operate laboratory and manufacturing equipment.

The Model Hardware Standard, or MHS, is intended to connect AI agents with programmable devices including microscopes and robotic arms. Anthropic said potential applications range from drug-discovery experiments to calibrating lasers used in quantum computing.

The Claude developer says the system could support autonomous workflows operating continuously with limited human involvement. MHS is designed to work with equipment that has a programmable interface and can enable communication between machines and AI agents over networks.

Anthropic is initially sharing the technology with partners for safety evaluation before a planned open-source release.