Anthropic Previews New Standard to Connect AI Agents with Lab and Factory Hardware

AI developer Anthropic announced on Thursday the launch of a research preview for its Model Hardware Standard (MHS), a new specification aimed at enabling AI agents to seamlessly operate physical equipment. The standard is designed to dramatically accelerate automation in laboratories and manufacturing by connecting AI models to devices like robotic arms, microscopes, and liquid handlers.

A New Standard for Physical Automation

Developed in collaboration with HHMI Janelia Research Campus, MHS provides a universal framework for AI-to-hardware communication. According to Anthropic, the standard can reduce hardware integration time from weeks or months to hours or minutes.

The system is designed to be model-agnostic, meaning it can work with any AI agent harness. It utilizes a standardized driver that translates simple commands for complex hardware, allowing different devices and AI agents to communicate over a network without requiring custom programming.

Broad Industry Adoption

The initiative has already garnered support from several major technology and life sciences companies that are integrating MHS into their platforms and equipment. Key partners include:

Ad

• Amazon Web Services (AWS), which will support MHS through its Strands Robots library for connecting AI to physical devices.
• Danaher, which is exploring MHS with Anthropic to enable smart instruments and scale biomedical research in autonomous laboratories.
• Universal Robots and Doosan Robotics, which are testing the standard with their robotic arms for industrial and automated quality assurance applications.
• Life sciences firms QIAGEN, Tecan, and Automata, which are adding MHS support to their respective lab automation and liquid handling platforms.

Path to Open Source and Safety

Anthropic stated that the research preview period will be used to collaborate with its launch partners on building robust safety evaluations. The company is focused on developing protections for the use of AI with physical equipment.

Following the preview, Anthropic plans to open-source the standard to encourage widespread adoption. The company also intends to release its findings to provide guidance on the safe deployment of AI-controlled physical systems. Anthropic is now inviting stakeholders across relevant industries to join a waitlist for the research preview.