The UK Atomic Energy Authority has spun engineering-automation technology into Singular Machines, creating a standalone company backed by investors including Oxford Science Enterprises and engineering group Arup.
The business has completed a strategic investment round involving the UK Innovation and Science Seed Fund, managed by Future Planet Capital, Oxford Science Enterprises, Arup, and Japan’s Miraisozo Investments. The value of the round has not been disclosed.
Arup has also signed a commercial agreement to use and test Singular Machines’ coEngen platform, giving the spinout both investment backing and an industrial customer as it moves beyond the research environment in which the technology was developed.
Singular Machines is building coEngen as an agentic-AI system for complex engineering projects. The platform is designed to automate parts of the coordination and analytical workload while maintaining traceability, assurance, and professional judgement around engineering decisions.
The technology grew out of fusion research, where projects involve closely interdependent mechanical, electrical, structural, scientific, and safety requirements.
Those conditions create a demanding environment for automation. A change to one element of a project can affect specifications elsewhere, while engineering decisions have to remain traceable to their underlying assumptions and constraints.
Adrian van Arkel, founder and chief executive of Singular Machines, said engineering projects “have become so complex that they are reaching the limits of what human coordination alone can handle.”
The company intends to apply the approach beyond fusion, targeting complex machines, facilities, and infrastructure where engineering teams have to manage large volumes of technical information and cross-disciplinary dependencies.
Arup’s involvement provides a route to test those claims against operating projects rather than limiting development to controlled demonstrations. The engineering group will provide technical, commercial, and strategic support while helping shape the platform through real-world workflows.
Singular Machines has also been recognised through Innovate UK’s Agentic AI Pioneers programme, winning the advanced-manufacturing category for coEngen. The system is designed around multiple AI agents working from a shared data model while keeping optimisation decisions traceable.
The spinout forms part of a broader effort by public research organisations to convert intellectual property and specialist technical knowledge into businesses capable of attracting private capital and commercial customers.
UKAEA’s 2026–2030 strategy places greater emphasis on creating economic value around fusion research, including technologies that can be applied in sectors outside energy. Fusion development draws on robotics, advanced materials, data, control systems, manufacturing, and artificial intelligence, giving some of the resulting technologies commercial uses beyond the original programme.
That offers a route to economic return from public research before commercial fusion power itself reaches the grid. A tool developed to solve an engineering problem inside a national laboratory can generate revenue, employment, and productivity gains elsewhere if it proves transferable.
Agentic AI is attracting particular attention because it extends beyond generating text or analysing a single prompt. Agent-based systems can break larger tasks into stages, call tools, retrieve information, and coordinate sequences of work, allowing deeper automation inside professional workflows.
Engineering places stricter demands on those systems than many administrative applications. Designs have to satisfy standards, technical constraints, budgets, safety requirements, and physical reality. A plausible output has limited value if the underlying calculation, assumption, or design change cannot be traced.
Adoption will therefore depend as much on governance and assurance as model performance. Engineering organisations need to decide which tasks can be delegated, how automated work is checked, where records are retained, and who remains professionally accountable.
Singular Machines is positioning coEngen around that boundary between automation and professional control rather than attempting to remove engineers from the process.
The presence of Arup as both investor and customer gives the company an opportunity to demonstrate whether the system can improve coordination on projects where multiple specialist teams need to work from consistent information.
The investment round also gives Singular Machines a broader shareholder base as it moves out of UKAEA and into the commercial market. Its next phase will depend on turning technology developed around high-assurance research into repeatable software that engineering and infrastructure organisations are prepared to deploy at scale.
You must be logged in to post a comment.