Databricks is expanding the Partner Well-Architected Framework (PWAF) to help partners build on, connect to, and share through Databricks with more up-to-date guidance for the AI era. The framework is designed to give partners architecture patterns, technical standards, and implementation guidance that move with the platform instead of staying stuck in static documents.
Key highlights
- AI-ready guidance for partners: PWAF gives partners a single architecture guide for core patterns like Built-On, Connected, and Data Collaboration, with guidance they can use directly when building Databricks-based products and integrations.
- New AI Partner Dev Kit: Databricks has added an AI Partner Dev Kit with task-focused skills for coding assistants, covering areas like authentication, telemetry, Unity Catalog, and connector patterns to help partners build against vetted standards faster.
- More pattern coverage: Since launch, PWAF has expanded guidance for areas like clean rooms, software-defined storage, Genie-related AI patterns, Lakebase integration, and MCP Marketplace validation.
- Firefly is now open source: Firefly, the reference implementation for Built-On partners, is available as open source and gives partners a starting point for areas like auth, enterprise security, embedded apps, and AI-powered experiences on Databricks.
- Built to keep evolving: Databricks positions PWAF as a living framework that keeps adding patterns, skills, and reference implementations as the platform and AI market continue to change.
๐ Read the full post here
