QuEra Automates Critical Quantum Computer Process with Anthropic's AI Agent

Quantum computing company QuEra has made significant strides in automating a critical process for its quantum computers. The breakthrough, achieved through collaboration with artificial intelligence (AI) developer Anthropic, involves the use of an AI agent to restore a crucial laser system in seconds. This development could potentially reduce the need for on-site specialists as larger and more complex quantum computers are deployed at customer facilities.

The key to this achievement lies in QuEra’s partnership with Anthropic, which has developed software that can automatically recover from disturbances affecting the laser system. The AI agent, named Claude, was tasked with writing a control program using the Model Hardware Standard (MHS), a framework designed for laboratory and manufacturing equipment operation within preset safety limits.

The results of this collaboration are promising, as QuEra’s controller successfully recovered the laser system in 695 out of 700 timed trials across seven fault types. Notably, Claude never falsely reported a successful recovery during any trial, demonstrating its reliability and accuracy.

QuEra has previously automated recovery from routine laser disruptions on its Aquila quantum computer, which boasts an uptime of over 99%. However, more complex failures have continued to require human judgment and intervention. The company’s vice president of quantum systems, Sergio H. Cantu, emphasized the significance of this development: ‘We built a solution using the Model Hardware Standard to fix that: the lock recovers itself in seconds, verified every time.’

The process involved assigning the problem to Claude through MHS, which allowed the AI agent to conduct experiments on a dedicated laser testbed, evaluate results, and revise its approach. Engineers determined Claude’s operating limits, reviewed its work, and established conditions required for success. The equipment also specified safety boundaries, interlocks, and emergency-stop procedures that the agent was required to follow.

The resulting controller handled recovery without an AI model making decisions while the system operated. In 700 timed trials covering seven types of faults, most were resolved in less than six seconds, with the most difficult taking about 10-14 seconds. QuEra noted that a specialist would typically require five to ten minutes to resolve comparable problems.

The testbed was exposed to ordinary foot traffic and environmental disturbances rather than controlled test conditions, demonstrating Claude’s ability to recover from naturally occurring disruptions without human assistance. The AI agent also improved the stability of the laser lock by reducing remaining noise by a factor of five and preventing system loss during unattended operation.

QuEra applied this approach to a laser operating at a second wavelength, with Claude determining settings during one overnight run. This development could streamline commissioning work for future installations, which would otherwise require weeks of hands-on effort from specialists. The company’s president, Takuya Kitagawa, highlighted the importance of autonomous maintenance: ‘A customer expects the entire computer, and thus every subsystem, to hold itself together without a specialist in the room.’

QuEra is working with Amazon Web Services on cloud delivery of its planned Libra system through Amazon Braket in 2028. The company also has partnerships with Hewlett Packard Enterprise for integration with high-performance computing systems and NVIDIA for accelerated computing.

The collaboration between QuEra, Anthropic, and the Howard Hughes Medical Institute’s Janelia Research Campus began as a research preview of MHS. This framework is still in development, but its safety design is being evaluated through limited access by application. The results from this research preview demonstrate the potential for AI agents to automate critical processes in quantum computing.

QuEra’s achievement marks significant progress towards reducing the need for on-site specialists as larger and more complex quantum computers are deployed at customer facilities. This development could have far-reaching implications for the industry, enabling customers to expect autonomous maintenance of their systems without requiring human intervention.