Physical AI automates welding jobs that could not be tackled before
1 September 2026 
Novarc Technologies, a Canadian company that specialises in applying physical AI to welding automation, has joined forces with the US arc welding equipment manufacturer, Miller Electric, to develop a system that brings robotic welding to applications that have traditionally been difficult to automate due to part variation, fit-up inconsistencies, tack welds and the challenges associated with large, complex fabrications.
Many fabrication jobs are currently beyond the reach of conventional automation because weld conditions are not always predictable. The jointly developed Copilot Builder with Blue iQ system uses adaptive intelligence to react continuously to weld variations in real time, helping fabricators to take on a wider variety of applications without needing extensive reprogramming or expensive fixtures.
“The value of automation is well established, but many fabrication applications have remained difficult to address with conventional systems where conditions aren’t always predictable,” explains Miller’s business unit director, Sam Harvey. “Fabricators need solutions that fit the realities of their operations, where part fit-up, joint conditions and production demands can change from one job to the next.
“With Copilot Builder with Blue iQ, the system can adapt to variation as it occurs, helping manufacturers to automate a broader range of welding applications, improve productivity and make better use of their skilled workforce while maintaining the flexibility their operations demand.”
“Fabricators supplying data centres, shipbuilding, energy infrastructure and heavy equipment are facing a growing shortage of skilled welders, programmers and challenges related to part variability in their robotic welding operations,” adds Novarc’s co-founder and CEO, Soroush Karimzadeh. “Harnessing our vast amount of welding data, we are excited to bring our physical AI platform to the Copilot Builder to solve these robotic welding challenges. NovAI, our AI-powered computer vision solution, performs real-time adaptive welding with no prior joint scanning. Through our partnership, welding automation can now be achieved across a broader range of welding applications.”
The Copilot Builder will be integrated with Novarc’s centralised enterprise welding intelligence platform called NovHub, which brings visibility to robotic and mechanised welding operations by connecting weld data, video, process parameters and production records. This extends the system’s capabilities by converting production data into actionable insights.
Key technologies used in the Copilot Builder system include:
- NovAI autonomy This uses machine vision, real-time adaptation and adaptive control to help compensate for gaps, misalignments and part variations during welding.
- Adaptability to variations This tackles fit-up variability, tack consumption and changing weld conditions without needing constant reprogramming.
- Flexible deployment A portable design helps to bring robotic welding into existing workspaces, rather than requiring parts to fit into traditional weld cells.
- Traceability with NovHub This captures and analyses weld data through a centralised platform that provides insight, traceability and optimisation.
The welding system can be mounted on horizontal, vertical or inverted surfaces to accommodate different workspaces and part geometries, and to integrate with existing workflows. Remote operation helps improve safety by allowing operators to work away from heat and fumes, while integrated data supports continuous process improvement over time.
Novarc specialises in bringing perception and cognition to arc-welding robots. Without visual intelligence these robots can make costly errors that lead to rework, scrap, rebuilds and downtime. Novarc’s physical AI platform combines real-time computer vision, adaptive control and welding intelligence, with over-the-air software updates, transforming robotic welders into continuously improving physical AI systems.
