What we do

From opportunity to real-world use

Use case and
system design

Find where machine learning, deep learning or agentic AI pays off for your business, and define the models, hardware, sensors and actuators it needs.

Human–machine
interaction

Help people understand what the system is doing, anticipate its actions and intervene when needed.

Simulation and
prototyping

Prototype algorithmic logic, SLM and LLM harnesses and data projections, and simulate the full use case before committing to development.

Safety and
real-world validation

Identify risks and requirements early, and plan testing under real operating conditions.

How to start

Physical AI
Concept Sprint

Turn a Physical AI opportunity into a working concept. In four weeks, we develop a focused behaviour prototype and assess its value, usability, technical feasibility, safety and economics, with your team involved from the start.

4 weeks Senior cross-disciplinary team
At the centre

Behaviour prototype

A tangible or simulated prototype of the critical behaviour, built to test the assumptions that matter most to your next decision.

  1. Value

    Is this task worth improving with AI?

  2. Interaction and trust

    Can people understand the behaviour and intervene when needed?

  3. Technology and data

    Can it sense and respond reliably, with the data and hardware available?

  4. Safety and regulation

    What requirements apply, and which risks need further testing?

  5. Economics and lifecycle

    Can it meet target cost, and what will operation, maintenance and upgrades require?

The outcome

Go, refine or stop

A recommendation you can defend, supported by evidence, open questions and next steps.

Could Physical AI improve
your product or process?

Bring us a product challenge or an early idea.
Together, we’ll identify where Physical AI could add value and what to test first.


Cases & insights

Perspectives on Physical AI, intelligent products and how people interact with them.


Questions about designing and developing Physical AI

Where could AI add value to our product?

We help identify tasks where sensing, interpreting information or adapting behaviour could improve the product or the way it is used. We assess the opportunity against user needs, technical feasibility and cost, including whether a simpler solution would do the job.


Can we work with you before choosing the technology?

Yes. We start with the task, the people using the product and its operating environment. Together, we clarify the behaviour required, the available data and the constraints. These guide decisions about sensing, hardware, software and AI.


Does Physical AI mean fully automating the product?

No. Physical AI can support people, automate specific tasks or allow a system to act independently within defined limits. The right approach depends on the task and the consequences of an error. We design how people understand the system’s actions, intervene and take control.

How do you handle confidential information when using AI?

We work with AI in controlled, secure environments. Before using your information, we agree which tools and data may be used, how access, storage and retention are managed, and whether data may be used for model training. We align this with your security requirements. Confidential or personal information is only shared with external AI services with your permission.

What can we learn from a Concept Sprint?

A focused prototype makes the proposed behaviour tangible. Alongside it, we assess value, interaction, technology and data, safety requirements, and economics. The outcome is a recommendation on what to develop, refine or stop, with the evidence gathered and the questions still to resolve.

How do you assess cost and long-term viability?

We look beyond the AI model to the hardware, sensing, computing, energy and integration it requires. We also consider maintenance, updates and service needs. This helps your team assess whether the proposed benefits justify the cost across the product’s life.