Keen AI has secured £355,985 in Alpha-phase funding from Ofgem's Strategic Innovation Fund to develop a shared artificial intelligence model for monitoring electricity grid assets. The project is backed by all three of Great Britain's transmission operators.
The system, known as Foundational Shared Model Operations, or FoSMo, is being developed with National Grid as lead partner. It also includes SP Energy Networks, SSEN Transmission, UK Power Networks and Electricity North West. The group aims to create a single model trained on anonymised data from different networks, which each operator can then adapt for its own use.
Operators currently develop separate computer-vision systems to inspect many of the same assets, including pylons, cables, insulators and fittings. That increases costs and can limit the number of examples available to train models to detect rare defects.
FoSMo is intended to standardise how visual data is collected and analysed across participating networks. By pooling anonymised datasets, operators would gain access to a broader base of information than any one company could assemble alone.
The funding follows a proof of concept carried out with National Grid. The Alpha phase expands the work to cover transmission and distribution operators across Britain's electricity system.
Shared model
The consortium estimates the shared approach could save about £22.6 million over five years if adopted across all participating operators. The savings would come from avoiding parallel development of similar models and reducing faults through earlier detection of asset problems.
Once fully adopted, the project is also projected to prevent about 85,000 customer interruptions a year and 5.2 million minutes of lost power. The focus is on identifying defects in overhead line components before they lead to service failures.
Britain's electricity network is physically interconnected, but the software used to inspect assets has largely been developed in isolation by individual operators. Because faults and damaged components are relatively uncommon, a single network owner may not gather enough examples to train a model with consistent accuracy across all asset types and conditions.
Keen AI has previously worked with electricity transmission and distribution customers in Britain and overseas. It has processed more than one billion images and will act as FoSMo's technical steward, including developing, maintaining and hosting the models in the UK.
Domestic control
Domestic hosting is a central part of the project. The companies involved say the model is designed so data remains under the control of each asset owner, with an emphasis on limiting data collection and long-term storage.
That reflects a wider policy focus on the electricity network as critical national infrastructure. The government has identified the grid as a national priority as it prepares for major expansion to support electrification and decarbonisation.
Supporters argue that a shared model built and hosted in Britain would allow operators to retain control over a key layer of digital infrastructure rather than rely on fragmented systems or overseas suppliers. They also see it as a way to provide a common monitoring framework as new overhead lines, substations and grid connections are added.
A wider rollout would come as network companies face growing pressure to manage ageing assets while expanding the system. Automation in inspection and condition assessment has become a greater focus as operators seek to identify equipment issues earlier and reduce the risk of faults.
Amjad Karim, Chief Executive Officer and Founder of Keen AI, set out the rationale for the approach. "The UK is investing tens of billions in its electricity network. We can either build the AI that manages it ourselves or hand that capability to someone else. FoSMo keeps it here, with AI developed collaboratively, data owned by the industry, and only getting better as the grid expands. When every major network operator shares what they know about their assets, we end up with something more robust than any of them could build alone. That's how the UK can future-proof a grid that's about to double in size," Karim said.
National Grid said the project could support network upgrades while reducing duplicated work across the sector. "We recognise the significant value that AI tools can bring to enhancing operational insight and efficiency, so we're very pleased to work with Keen AI on this project. By pooling data and expertise, the model has the potential to support our upgrades of the grid while also making considerable cost savings across the industry," Ward said.