Agentic AI Enhanced Software V&V

We have built an agentic assistance system built on a single knowledge graph. Every requirement, design element, test case and test log is a typed node, and every trace is a typed link that carries the source it was derived from. Specialised agents work down the V-model. They extract atomic, verifiable requirements from specifications, standards and legacy documents. They then draft test case specifications with preconditions, steps, expected results, and negative and boundary cases. Finally, they generate executable test scripts for your SIL/HIL environment. Every draft cites the exact passage it came from. An evaluator agent checks each answer against its sources before an engineer sees it.

FEV etamax already delivered this solution for ground segment software validation. The MVP was reviewed by the institutional customer and measured against the manual baseline, leading, amongst others, to 50–80% less time spent drafting functional test cases and impact analysis in under one minute.

The same platform can be used for onboard software or other domains such as railway and automotive.