ANYbotics unveils autonomous workforce vision for process industries, conducting hundreds of thousands of inspections a month

ANYbotics on September 9, 2026, published its “ANYbotics Manifesto,” a vision for an autonomous workforce for process industries. The company said its field robots conduct hundreds of thousands of autonomous inspections each month. The target sites include plants in the power, chemical, metal and cement industries. The company said it provides three integrated layers: Physical AI, Industrial AI and Workflow Integration.

ANYbotics is presenting an operating model built around feeding inspection data into maintenance work, rather than robots alone. As stated in the announcement, the design spans DCS, historians, CMMS and ERP systems, indicating that the ability to connect with existing plant systems is a condition for adoption. The company also targets power, chemical, metal and cement plants and says it uses models for individual equipment families. This suggests adoption may vary not by industry as a whole, but by equipment configuration and maintenance structure.

What is new in this announcement is that ANYbotics has repositioned itself not simply as an inspection-robotics company, but as a provider of an “autonomous workforce” for process industries. The figures it highlights — hundreds of thousands of autonomous inspections a month, 99% stable missions and 95% autonomous operation — support its claim that it has moved beyond isolated demonstrations and into the operational phase. At the same time, this is less a sudden new direction than a continuation of a trajectory that began with the development of an early electric legged robot by ETH Zurich in 2010, continued with the TotalEnergies Robotics Challenge in 2017 and the application of Physical AI in 2018, and has involved ANYmal since 2021. The manifesto can be read as elevating the discussion from individual capabilities to an overall design that connects field inspections with anomaly detection, maintenance workflows and ERP integration. In other words, the focus is shifting from competition over robot performance to operational competition over how collected data connects to equipment-maintenance decisions. There are three implications for the industry. First, the company’s stated integration with DCS, historians, CMMS and ERP shows that the success of robot deployment depends not only on mobility in the field, but also on the ability to implement the technology in existing plant systems. Second, modeling across equipment families such as pumps, compressors, heat exchangers and transformers points to a design aimed at deployment across equipment categories, rather than one-off projects at individual plants. Third, by naming Siemens Energy, GE Vernova, Yokogawa, SLB and Kanadevia Inova, the company reinforces the expectation that inspection robots will be integrated into existing ecosystems for control, maintenance and industrial software. Open questions include how many ANYmal robots are conducting the hundreds of thousands of monthly inspections, utilization rates by customer, the accuracy achieved for each equipment family, and how much the Adoption Accelerator contributes to sustained adoption at individual companies. For now, the focus is less on the robots’ performance itself than on how far operational data can be accumulated and integrated into maintenance work in a repeatable way.

The company said ANYmal can autonomously navigate stairs, grating and hazardous terrain, achieving 99% stable missions and 95% autonomous operation. It also said the robot is equipped with thermal, acoustic, ultrasonic and gas sensing. On the data side, Industrial AI runs models tailored to equipment groups such as pumps, compressors, heat exchangers and transformers. Workflow Integration is designed to connect with DCS, historians, CMMS and ERP systems, and the company cited close collaboration with Siemens Energy, GE Vernova, Yokogawa, SLB and Kanadevia Inova. ANYbotics also noted that its co-founders and others at ETH Zurich developed an early electric legged robot in 2010, reached the finals of the TotalEnergies Robotics Challenge in 2017, and that ETH Zurich announced the world’s first application of Physical AI for legged robots in 2018.

Yokogawa was named as a Workflow Integration collaborator. For Japanese companies, a key issue is likely to be how to connect robotic inspection with existing plant control and maintenance-management systems. The equipment targeted — pumps, compressors, heat exchangers and transformers — also overlaps with established areas of Japan’s industrial equipment and maintenance software sectors.