Software and AI Safety Assurance

A process-driven audit of software and AI safety assurance, without writing or testing the client’s code.

What we do

Software and AI can move safety decisions into development processes, data boundaries and behaviour that is not fully specified in advance. We audit the assurance process around those systems rather than writing or testing code.

Our review covers the software development lifecycle, safety planning, verification evidence, fault-tolerance logic, tool and data controls, and the boundaries of algorithmic capability. We assess whether the development and assurance approach is proportionate to the claimed safety function and applicable standard.

Where complex testing, simulation or model execution is needed, the client provides the licensed environment or raw model data. Our role is to define the assurance question, audit the process and make the basis for the conclusion clear.

Who this is for

Nuclear, defence, aerospace, rail, automotive, maritime, medical device and industrial organisations introducing software, autonomy or AI into safety-relevant systems.

Related standards

IEC 61508, ISO 26262, DO-178C and DO-333, EN 50128 and EN 50126, Emerging AI assurance guidance