What I am building
I treat TaricAI as a planned use case for SovereignFlow. The idea concerns tariff classifications, BTI/WIT decisions, and source documents. Classification rules, importers, and source presentation belong in the domain application, while the RAG engine supplies shared processing mechanics. For now it is a concept, without an implementation of a finished assistant.
Why a separate domain application
Tariffs and classification decisions require their own importers, source history, and interpretation rules. I do not want to embed that domain logic in SovereignFlow. TaricAI is a way to test whether a neutral engine can support a clearly defined application domain.
Project scope
- Proposed domain: TARIC and BTI/WIT
- Planned retrieval of similar classifications and supporting documents
- Separation of domain logic from SovereignFlow