This paper presents substrate-rooted identity as a systems-engineering framework for establishing identity, attestation, and provenance through execution behavior rather than static metadata. The approach addresses modern engineered systems that incorporate AI models, embedded controllers, firmware modules, and cyber-physical components whose behavior changes through fine-tuning, environmental interaction, updates, and nondeterministic execution. The framework defines the Engram Signature, behavioral attestation, execution-rooted provenance, and lifecycle-aligned identity invariants as architectural constructs for tracking continuity, detecting behavioral drift, and supporting lineage formation across heterogeneous systems. By modeling identity as a stable region in execution space, the method complements trusted-computing mechanisms and strengthens governance, verification, and trust architectures. The paper outlines the assumptions and boundary conditions under which substrate-rooted identity operates and provides an illustrative case study demonstrating applicability to real-world engineered systems. The contribution is conceptual and architectural, offering a lifecycle-oriented perspective on identity, provenance, and execution-based trust in distributed and software-intensive systems.