Embedding AI/BI dashboards in customer-facing applications is straightforward, but enforcing precise data access per viewer requires a robust authorization framework. A combined pattern of scoped tokens, row-level filters, and synchronized group entitlements on Databricks ensures secure, consistent views without proliferating dashboard copies.
- Use scoped tokens with __aibi_external_value to enforce per-viewer row filters
- Centralize access rules in entitlements table shared by embedded dashboards and direct SQL queries
- Run queries as service principal to secure credentials and simplify data permission management
Infrastructure signal
The new pattern leverages Databricks capabilities such as Unity Catalog’s row filters and column masking combined with externally minted scoped tokens containing viewer-specific claims. This integration shifts key authorization decisions upstream into token generation and entitlements management, enabling a unified security model for embedded dashboard access and native SQL querying.
A trusted backend service principal mints scoped tokens embedding viewer identity and access scope, enforcing data segmentation without relying on per-viewer dashboard duplication. The architecture isolates service principal credentials in a secrets manager to avoid exposure. Embedding utilizes the __aibi_external_value parameter in SQL queries, making queries inherently aware of viewer context while running under a single service principal identity.
Developer impact
Developers gain a streamlined workflow by consolidating data access rules in a single entitlements table rather than distributing filters across multiple dashboards or queries. This reduces deployment complexity, risk of filter drift or inconsistency, and testing overhead. Since the authorization information travels securely inside the token, client-side code remains simplified, focusing only on receiving and rendering dashboards.
Earlier concerns about maintaining multiple dashboards per customer or manually editing filter conditions for each viewer dissipate. Dev teams now can audit and update scopes centrally with immediate downstream effects. Additionally, embedding dashboards under a service principal account simplifies permission management by decoupling user identities from query execution context, enhancing maintainability and scalability.
What teams should watch
Cloud infrastructure and data platform teams need to ensure secure management of service principal credentials, integrating secrets management and token minting procedures into CI/CD pipelines to maintain security posture. Data engineers should align entitlements schema design with business requirements for partner and group segmentation to support scalable permissions enforcement.
Security and compliance teams should verify that default-deny rules are effectively enforced by testing unknown or malformed scope claims that result in zero data returns. Analytics product owners must coordinate with development to confirm that dashboards consistently enforce intended data visibility without duplication or performance degradation. Monitoring tools should track usage patterns of embedded dashboards and token validation metrics for anomaly detection.