A portfolio-data API typically hosts normalized positions and transactions so your apps query a remote source of truth. That is the right tool when you want managed sync, multi-app reads, and accept the vendor as a data processor with retention.
Open Portfolio sells an operated BYOC inference boundary: edge ingestion, buyer IdP/storage, bounded context, and stateless inference. The OSS importer proves parsing; Pocket proves chaos under real CSVs. We do not position as a drop-in replacement for every portfolio-data API.
Choose the API when you need hosted holdings as product infrastructure. Choose the boundary when AI or analytics must not create a new warehouse of client ledgers. Many stacks will use both — API for operational data, boundary for AI diligence.