What it is not

Scope Boundaries

To prevent misunderstanding, the following clarifications define what SERAPH is not.

• It is not a vector database. Similarity search is one of its mechanisms, not its purpose, and it is not restricted to text.
• It is not a traditional curated knowledge graph. Any structure may be declared, but no one is required to design a schema before storing anything, and the declared layer never substitutes for the emergent layer underneath.
• It is not a language model or an AI system. No inference occurs inside the substrate. Intelligence is implemented in a separate layer above it.
• It is not tied to any single modality. Text is the most common use case, not the only one.
• It does not claim that the content it records is true. It guarantees that the content it records, and the claims made about that content, are recorded, ordered, and unaltered. This is epistemic honesty about what was captured, not a verdict on whether what was captured was correct.
• Gov mode does not replace an accreditation authority. The enterprise tier enforces compartmented access control and per-operator signing at the substrate level. It does not perform user authentication, certificate chain validation, or clearance adjudication. Those responsibilities belong to the application layer and the customer's PKI infrastructure. SERAPH enforces the clearance context it is given; it does not originate it.