Amazon SageMaker Unified Studio adds custom asset types to the catalog in IAM-based domains
Catalog any asset format—PowerBI dashboards, medical images, PDFs—inside SageMaker Unified Studio with governed discovery and subscription workflows.
View original announcement →Visual Summary
graph TD
A{{Custom Asset Types}}:::announced
B((Domain Admin)):::external
C([Schema & Metadata Forms]):::feature
D([Asset Enrichment]):::feature
E([Governed Subscriptions]):::feature
F(SageMaker Unified Studio):::compute
G(Amazon S3):::storage
H((Domain Members)):::external
I([Catalog Discovery]):::feature
J(AWS IAM):::compute
B ==>|"defines"| A
A -->|"validates against"| C
A -->|"publishes to"| F
F -->|"enables"| I
H -->|"searches"| I
I -->|"requests"| E
A -.->|"references"| G
A -->|"enriched with"| D
F -->|"secured by"| J
classDef announced fill:#ff9900,stroke:#ec7211,color:#fff,font-weight:bold
classDef compute fill:#e3f2fd,stroke:#1565c0,color:#1565c0
classDef storage fill:#e8f5e9,stroke:#2e7d32,color:#2e7d32
classDef feature fill:#fff3e0,stroke:#e65100,color:#e65100
classDef external fill:#f5f5f5,stroke:#616161,color:#616161
What's New
Amazon SageMaker Unified Studio now supports custom asset types for IAM-based domains, enabling domain administrators to catalog virtually any asset format—from medical imaging files and PowerBI dashboards to third-party PDF reports—directly within the SageMaker catalog. Administrators can define custom asset types with structured metadata schemas, then create, enrich, and publish individual assets for organization-wide discovery and governed subscription workflows. This capability eliminates the need for separate cataloging tools by bringing all organizational assets, regardless of format or origin, into a single unified discovery experience.
How It Works
- Custom asset type creation: Domain administrators navigate to Domain Management > Custom Asset Types and define a new type with a unique name (1–128 characters), an optional description (up to 2,048 characters), and optional metadata form types that establish required or optional fields each asset of that type must carry.
- Schema versioning and validation: Asset types in SageMaker Unified Studio are versioned; when individual assets are created from a type, they are validated against the schema defined by that type, and creation fails if the structure is invalid—ensuring catalog consistency.
- Asset creation and enrichment: Once a custom asset type exists, individual assets are instantiated from it and can be enriched with glossary terms and README documentation to provide business context consumable by both human users and AI agents.
- Publishing and discovery: Enriched assets are published to the catalog, where any domain member can search by asset name, type, or glossary term through the standard SageMaker Unified Studio catalog interface.
- Governed subscription workflow: Discovered assets can be subscribed to via the same governed request-and-approval workflow used for all native catalog assets; however, subscription fulfillment itself must be managed outside SageMaker Unified Studio by the asset owner.
- IAM-based domain scope: This feature is currently scoped to IAM-based domains within SageMaker Unified Studio, leveraging IAM for identity and access control across the catalog operations.
Why It's Important
- Eliminates catalog fragmentation: Organizations typically maintain multiple disparate catalogs for different asset types (data lakes, BI tools, document repositories); custom asset types consolidate all of these into a single searchable catalog, reducing tool sprawl and context switching.
- Extends governance to non-native assets: Assets from third-party platforms or proprietary formats previously fell outside governed discovery workflows; now they benefit from the same metadata curation, glossary tagging, and subscription governance as native SageMaker assets.
- AI-agent readiness: Enriching assets with glossary terms and README documentation makes them machine-readable, enabling AI agents and LLM-based workflows to discover and reason about organizational assets programmatically.
- Accelerates data democratization: Self-service discovery and subscription requests mean data consumers no longer need to contact asset owners directly or navigate separate systems, reducing time-to-access for critical business assets.
- Supports regulated industries: Use cases like cataloging medical imaging files (DICOM, NIfTI) or compliance documents become tractable within a governed AWS-native environment, which is significant for healthcare, finance, and life sciences organizations.
How It's Different
- Versus native SageMaker asset types: Native types (tables, ML models, dashboards) have built-in fulfillment and integration; custom asset types are format-agnostic but require the owner to handle subscription fulfillment externally, trading automation for flexibility.
- Versus AWS Glue Data Catalog: Glue is optimized for structured/semi-structured data assets with automatic schema inference; custom asset types in SageMaker Unified Studio support arbitrary formats with human-defined schemas and richer business-context metadata (glossary terms, READMEs).
- Versus third-party data catalogs (Collibra, Alation): Those tools offer broad asset cataloging but are external to the AWS ML/data workflow; SageMaker Unified Studio integrates cataloging directly into the environment where data scientists and engineers already work, reducing context switching.
- Versus Amazon DataZone custom asset types: SageMaker Unified Studio is built on DataZone foundations but surfaces this capability within the unified ML/analytics IDE experience, making it accessible without navigating a separate service console.
- Schema versioning: The versioned schema model ensures backward compatibility and audit traceability as asset type definitions evolve, a feature not universally present in lightweight cataloging solutions.
When to Prefer It
- Multi-format organizational catalogs: When your organization manages assets across heterogeneous formats (proprietary files, third-party exports, legacy system outputs) and needs a single governed catalog rather than format-specific silos.
- Regulated data environments: When cataloging sensitive assets (medical images, financial reports, legal documents) that require governed access workflows, audit trails, and metadata standardization within an AWS-native environment.
- AI/ML pipeline enrichment: When building agentic or RAG-based workflows that need to discover and consume diverse organizational assets programmatically, and where glossary-enriched metadata improves retrieval accuracy.
- BI and analytics asset governance: When revenue dashboards, PowerBI reports, or Tableau workbooks need to be discoverable and subscribable by data consumers without granting direct access to the BI platform itself.
- Centralized data marketplace initiatives: When standing up an internal data marketplace where producers publish assets and consumers request access through a self-service workflow, regardless of where the underlying data physically resides.
- IAM-based domain deployments: When your SageMaker Unified Studio deployment uses IAM-based identity (as opposed to IAM Identity Center-based domains), and you need catalog extensibility within that architecture.
Availability
- GA status: Generally available as of July 9, 2026; this is not a preview feature.
- Domain type: Available exclusively for IAM-based domains in Amazon SageMaker Unified Studio; IAM Identity Center-based domain support is not mentioned in this release.
- Regional availability: Available in all AWS Regions where Amazon SageMaker Unified Studio is currently supported; no region-specific exclusions noted.
- Pricing: No separate pricing announced for custom asset types specifically; costs are expected to follow standard SageMaker Unified Studio and Amazon DataZone catalog pricing—consult the AWS pricing page for current rates.
- Fulfillment limitation: Subscription fulfillment for custom asset types must be handled outside of SageMaker Unified Studio by the asset owner; the platform manages discovery and subscription requests only, not data delivery.
- Metadata form optionality: Metadata forms attached to custom asset types can be required or optional, but asset creation will fail validation if required fields are not populated.
