Anthropic’s new AI system lets lab machines talk to each other and run experiments

Anthropic's Model Hardware Standard (MHS) lets different laboratory instruments communicate and work together through AI agents. The system can help automate experiments by enabling AI to control robotic arms, liquid-handling machines and other lab equipment.

A new artificial intelligence platform is making it easier for lab equipment to communicate with one another, which could help speed up and streamline research automation. Anthropic's Model Hardware Standard (MHS), created in collaboration with the Janelia Research Campus at the Howard Hughes Medical Institute, enables a wide range of lab tools to communicate with one another and can also enable AI agents to run and coordinate experiments, as reported by Nature.

Researchers at Carnegie Mellon University in Pittsburgh, Pennsylvania, are testing the system, the report said. The aim is to address a common challenge in research laboratories: machines from different manufacturers use different software and programming languages, making them difficult to connect. MHS supports a layer of software that enables instruments to communicate with one another, without requiring individual integrations for each instrument.

MHS enables AI agents to control lab equipment

MHS acts as a software interface between a laboratory computer and its instruments. Researchers can connect a programmable device and describe its capabilities to an AI agent using plain language. Other technical documentation, such as equipment manuals, can also be provided to help the system understand the instrument.

Once the equipment is connected, the researchers can describe an experiment, and the AI agent can manage the necessary equipment. In one demonstration reported in Nature, a robotic arm positioned a multi-well plate between instruments, which then loaded a single plate into a liquid-handling machine, followed by a second instrument for analysis.

Pragya Singha Roy