Systems we’re building toward.
These are the directions we are working on. They are not finished products. Prototypes and demos will show up here as they exist.
Aviation · AI · Optimisation
Predictive MRO planning under real-world constraints
We’re exploring an MRO planning system that combines forecasting with dynamic constraint solving. The objective is to predict maintenance demand while continuously accounting for parts availability, maintenance requirements, BOM structures, parts, labour, scheduling constraints, and other operational realities.
- Approach
- Rather than treating forecasting and planning as separate problems, predicted demand becomes an input to a constraint-based planning layer that can adapt as conditions change.
- Objective
- Move beyond forecasting what maintenance may be required toward generating practical maintenance plans that remain viable as operational constraints change.
- Demo direction
- Early demonstrations will focus on synthetic or representative maintenance data, showing how forecasts interact with constraints and how a plan changes when availability, parts, or other conditions change.
Forecasting · constraint programming · optimisation · MRO systems · operations research
Robotics · Simulation · Engineering
Simulation and emulation for robotics
We’re exploring a software environment for simulating and emulating robotic systems before they are deployed on physical hardware. The focus is on environments where behaviours, control systems, sensor inputs, and failure scenarios can be developed and tested.
- Approach
- The longer-term goal is to create a bridge between simulation and physical systems, allowing software and control logic to move from controlled virtual environments toward increasingly realistic hardware testing.
- Objective
- Reduce the cost and risk of robotics development by moving more experimentation, validation, and failure analysis into software.
- Demo direction
- Initial demonstrations will focus on simulated environments, robotic behaviours, sensor models, and controlled failure scenarios before progressing toward hardware integration.
Robotics · simulation · emulation · control systems · synthetic environments
Conservation · Spatial AI · Sensing
Long-range depth mapping without conventional cameras
We’re investigating whether radio-based sensing, including Wi-Fi signals, can be used to infer depth, movement, and spatial structure over distances where conventional cameras become impractical.
- Approach
- The research will explore how radio-frequency measurements can be transformed into spatial representations and whether learned models can extract useful information about environments and living organisms from those signals.
- Objective
- Develop non-visual sensing technologies with potential applications in animal conservation, ecological monitoring, and biological research.
- Demo direction
- Early work will focus on controlled sensing experiments and small-scale depth or movement reconstruction. Longer-term research could investigate larger-range sensing and environments where conventional cameras are difficult to deploy, including potential marine applications.
- Applications
- Note
- This direction combines machine learning with sensor and hardware experimentation. Long-range and underwater applications are research goals rather than established capabilities, and would require dedicated hardware, experiments, and suitable datasets.
Wi-Fi sensing · RF signals · depth estimation · spatial AI · representation learning · specialised hardware
Voice · AI · Human-in-the-loop
Live voice systems designed around human operators
We’re exploring an alternative approach to automated call centre systems: use AI to improve the workflow around the human operator rather than attempting to replace them. Part of that is quality of life — reducing the grind and shielding operators from the more uncomfortable parts of the role where technology can take the first hit.
- Approach
- A caller enters through a speech interface. Speech is transcribed, relevant information is extracted, and the case is routed into a workflow where a human handles the substantive interaction. AI can provide context, classification, retrieval, summarisation, and other targeted assistance — including absorbing repetitive or confrontational front-line load before it reaches the operator.
- Objective
- Improve operator quality of life by cutting repetitive operational work, giving people better context for the cases that need them, and keeping humans responsible for complex decisions without leaving them stuck in the most draining parts of the job.
- Demo direction
- Early demonstrations will focus on a complete but deliberately narrow workflow: caller → speech recognition → structured case information → human operator context, with AI introduced only where it provides a measurable benefit.
- Applications
Speech-to-text · voice infrastructure · workflow automation · information extraction · fine-tuned models · human-in-the-loop systems
Active research, not a client portfolio. Funding would go toward the next slice of work on a direction — experiments, a demo, or hardware where that is required.