Automated Hardware Validation Test Plan Generation for Large Scale AI Datacenter Platforms Using a Generative AI Multi-Agents Architecture

This paper presents a generative-AI multi-agent architecture for producing structured hardware validation test plans from self-healing validation documents and component Bills of Material. An ingestion agent normalizes heterogeneous inputs, a classification agent maps components to functional domains, and a generation agent combines failure modes with component context to create test cases, including gap-filling and edge scenarios. On two production platforms, the framework reportedly expanded coverage by 74.2% and 51.4% versus manual baselines, reduced authoring from days to hours, and provided traceability from every test case to its source specification. The output is designed for direct import into internal validation software.