[FEATURE] AI assisted curation of assets in domains and subdomains #98
Description
Is your feature request related to a problem? Please describe.
Early-stage customers experience friction when establishing their initial domain structure in Atlan. Specifically:
- Customers lack conceptual understanding of domains/subdomains and practical guidance on asset organization
Describe the solution you'd like
An AI-powered domain discovery and suggestion system that:
- Analyzes existing assets using our vector embeddings to identify natural groupings
- Suggests initial domain/subdomain structures based on asset metadata, naming patterns, and relationships
- Provides educational context explaining why certain domains are suggested
- Offers ability to accept, modify, or reject suggestions before implementation
- Includes templates for common industry patterns as a starting point (needs discovery with PM)
Technical Details
Proposed API changes:
- New methods in pyatlan for domain/subdomain creation and management
Impact on existing functionality:
- Backwards compatible - manual domain creation remains unchanged
- Enhanced discovery APIs to support domain-aware searching
Required dependencies:
- pyatlan domain management capabilities (prerequisite)
- Integration with existing vector embedding infrastructure
- MCP tools enhancement for domain curation workflows
Describe alternatives you've considered
Additional context
Use cases:
- New customer onboarding acceleration
- Existing customers reorganizing their data landscape
- Migration from other data catalogs
Success metrics:
- Time to first domain creation
- Domain adoption rate in C2/C3 segments
- Quality score of AI-suggested vs manually created domains
Implementation considerations:
- Need to validate AI suggestions don't reinforce poor data practices
- Must handle edge cases where assets don't fit clean domain boundaries
- Consider phased rollout: templates → guided setup → full AI suggestions