Changes with no impact on existing DDL/API functions

AI Copilot

• Copilot on the Semantic model page: Copilot now works directly on the Semantic model page — create semantic models and switch between them, add tables and views, and set business names, descriptions, and column semantic properties (semantic type, synonyms, sample values, exposure) per model. Copilot also navigates to and from the Semantic model page when a task belongs there, and tells you honestly when something can't be done instead of reporting false success
• MCP write access: MCP now supports write operations in addition to read — AI agents can create projects and branches through MCP, gated by a new workspace-level kill-switch that admins toggle in account settings, with an audit trail of write activity for governance
• Bring Your Own Model (BYOM): per-user cost tracking and balance-neutral metering for BYOM companies, so users can see their own usage without it affecting SqlDBM credits
• Object-level skills: Copilot Skills can now be scoped to specific object types (tables, views, columns) and invoked directly from an object
• A ready-made Copilot Skills template for semantic models, to help teams build AI workflows around their semantic layer faster
• An inline AI "spark" button to generate SQL expressions in semantic models
• Multi-step Copilot changes are now grouped into a single undo, so a change reverts as one action instead of step by step
• (Fix) More reliable Copilot operations: deletions, bulk changes across columns, Data Governance fields, schemas, databases, and relationship types, auto-layout across multiple diagrams, and diagram availability while Copilot is in fullscreen mode

Databricks & BigQuery

• Tags and labels: create, edit, and bulk-assign key/value tags on objects (Databricks tags and BigQuery labels), import them from XLSX, filter and search by tag, and see them on the diagram, in the Database Explorer, documentation, reports, and compare

Semantic Modeling

• Push semantic models to Git: Push-to-Git now includes semantic model YAML files alongside DDL, with a dedicated repo folder and a semantic model file tree in the Push preview. For Snowflake, semantic views are also pushed as source DDL. Semantic push works from branch merges too, and empty semantic YAML files are filtered out automatically

dbt

• dbt Fusion support: updated the Forward Engineering screen and added the new loaded_at_q field
• (Fix) dbt export now rejects non-YAML properties with a clear error and produces cleaner Fusion config blocks

Reverse Engineering

• (Fix) Direct Connect now honors the "Discard names on upload" option
• (Fix) Constraint and index column resolution during Reverse Engineering now correctly handles nested STRUCT columns
• (Fix) Databricks MERGE parsing no longer reports false-positive duplicate columns when parsing multiple tables in one batch
• (Fix) Improved Erwin and dbt converter reliability, including correct BigQuery column-name handling

Excel

• (Fix) Excel import no longer fails when a virtual relationship has no column pairs specified

OpenAPI

• (Fix) DDL revisions created via the API no longer reset project-level naming conventions and dbt properties — these settings are now preserved across API-created revisions

Erwin Import

• (Fix) Erwin XML imports into Logical projects now correctly populate logical names for entities and attributes