Does Bildstak copy my data or query it in place?

Question

Accepted Answer

Bildstak uses a federation model: it queries your data in the systems where it lives rather than copying it into a central store. For deployments that require it, Bildstak can also run fully air-gapped or in your own VPC, keeping data entirely within your control. The distinction between copying data and federating it is important for both security and data-quality reasons.

Copying data means extracting it from source systems, transforming it, and loading it into a central store (a data warehouse or data lake). The copy is immediately out of date; keeping it current requires an ongoing ETL pipeline; and your data now exists in an additional location, raising storage, compliance and access-control questions.

Federation means querying data where it lives. Bildstak connects to your Primavera P6 schedule, your Procore RFI register, your SAP cost ledger — and runs queries against them in place. The results are joined and returned to the user, but nothing is permanently duplicated.

What Bildstak actually does:
Bildstak uses a federated query model. To make queries fast and enable cross-source joining, it may maintain a lightweight index of identifiers and metadata — but it does not replicate your full dataset. IFC models are read and indexed locally; API sources are queried live with metadata cached for performance.

Deployment implications:
- Cloud deployment: queries run from Bildstak's infrastructure to your connected systems via their APIs.
- Customer-VPC or on-premise: the Bildstak federation layer runs inside your environment, so queries never traverse to external infrastructure.
- Air-gapped: the same applies, with no outbound connectivity required.
- Desktop app: IFC and local files are queried entirely on the workstation.

In every case, your data does not move to a third-party store, and Bildstak does not use it to train any model.

Bildstak copie-t-il mes données ou les interroge-t-il en place ?

Related questions