Diligent Robotics rolls out Moxi 2.0 after five years of hospital deployments
Moxi 2.0 from Diligent Robotics is rolling out to health system customers across the U.S.
After five years of operation in over 25 hospitals, the company has shaped the next-generation platform accordingly.
Diligent is also launching its learning flywheel with the new robot.
The proprietary world model learns from fleet experience and improves with each deployment.
CEO Andrea Thomaz said Moxi 2.0 is built with the compute, sensors and models to reason about complex hospital environments in real time.
Initial deployments include Endeavor Health Edward Hospital, Providence Saint John’s Health Center, and Children’s Hospital Los Angeles.
This is Diligent’s first major announcement since its acquisition by Serve Robotics in January 2026.
The company, founded in 2017, has deployed Moxi in more than 25 U.S. hospitals to assist nurses.
Moxi 2.0 offers faster, more confident deliveries and longer operating hours.
It uses NVIDIA Isaac Sim and Cosmos open world models, with upgraded NVIDIA A2000 compute that perceives surroundings 10-15 times faster.
Key upgrades include 10x onboard compute, a new Robotic World Model, up to 18 hours of runtime per day with 30% faster charging, improved autonomy and recovery, upgraded cameras/sensors/storage, and redesigned handles based on nurse feedback.
Omkar Kulkarni of Children’s Hospital Los Angeles said Moxi has completed over 40,000 deliveries, saving staff 16,000 hours of work.
The hospital expanded its fleet from two to three robots, with utilization growing more than 10% in Q2.
Questions remain about Moxi 2.0's base.
Diligent designed it modularly to continue using Fetch Robotics' base or switch later.
The robot's learning flywheel captures richer signals from hospital environments, feeding into cloud training infrastructure developed with AWS.
T-Mobile for Business ensures reliable network access, with all safety and autonomy behaviors running onboard.
Moxi works even without Wi-Fi, using cellular fallback when needed.
Diligent said this is the beginning of what its learning flywheel makes possible, with continuous improvement from deployment data.