Ready to Work Together?
Sustainable Electric Vehicles
Accelerating Electric Vehicle Charging Optimisation using LLMs
Efficient energy utilisation and cost optimisation for electric vehicle (EV) charging require the seamless integration of supply and demand data from multiple sources. These include industry stakeholders such as smart microgrid operators, charging station providers, and end users. Any effective solution must account for user preferences, dynamic pricing structures, and the operational constraints of the grid.
Intractābilis has developed a large language model (LLM)-driven dynamic demand management system designed to address these challenges. The system reduces the need for deep domain expertise, accelerates deployment, and simplifies customisation for diverse operating environments.
By applying retrieval-augmented generation techniques, the solution leverages existing optimisation engines rather than rebuilding them for every new configuration. This approach promotes reusability, enhances maintainability, and shortens development cycles.
Our system enables energy companies to deploy advanced optimisation capabilities more rapidly, reduce operational costs, and deliver an improved experience to EV users. The benefits are particularly strong in residential contexts and distributed energy environments, where flexible, data-driven charging strategies can significantly improve efficiency and user satisfaction.
Quantum Advantage in Smart EV Infrastructure Management
Electric vehicle (EV) charging requires a finite duration, with the starting time typically scheduled in advance by users. Each EV must be charged within a specific time window and is assigned to a group, which may represent a corporate fleet or alternative charging intervals for a single vehicle.
The introduction of pre-booking systems in smart charging enables users to select subsets of possible charging intervals. This provides terminal operators with greater flexibility in allocating EVs to charging points, allowing more efficient utilisation of available infrastructure. The operational challenge is to serve the maximum number of EVs from distinct groups at a single terminal, given a set of time intervals partitioned into groups.
This optimisation task can be framed as a Group Interval Scheduling Problem (ISP), a well-known NP-hard problem in classical computing. To address this, Intractābilis has developed a Quantum Simulation-Based Optimisation (QuSO) formulation that defines a superposition of feasible configurations and evaluates them simultaneously using the Quantum Approximate Optimisation Algorithm (QAOA). By applying the Quantum Singular Value Transform (QSVT), the solution reduces runtime complexity, achieving a significant acceleration in simulation-based optimisation.
The solution incorporates realistic hardware constraints and avoids problematic regions with vanishing energy gaps, ensuring the approach is viable for real-world applications. Experimental results highlight the potential of quantum hardware to deliver a meaningful advantage in producing high-quality approximate solutions for complex, large-scale EV charging optimisation challenges, offering clear commercial and operational benefits in the evolving clean energy and transport sectors.
Our solutions transform how commercial, industrial, and municipal stakeholders manage EV integration—enabling reliable decarbonisation without operational compromise. From EV fleet operators to city energy planners, GridSense AI offers the intelligence needed to coordinate complex, real-time energy ecosystems with confidence. It stands apart from conventional DSM approaches by offering not only predictive optimisation but also the generative reasoning needed to navigate the full complexity of distributed, renewable-rich grids.
© 2025 Intractābilis