9.7.11: Security groups, and Simba builds everything
A layered permission model enforced on every channel, an agent that creates any artifact, and parallel extraction for large tables.
The largest release of the year so far.
Security groups and fine-grained permissions. Global and tenant user groups, per-subsystem permissions covering View, Create, Edit, Run/Use, and Delete, artifact access through a Tag View, our tag-based artifact view, enforce and shadow and off rollout modes, an Effective access report, and administrator-only View as user. Roles remain the ceiling, and every layer is enforced across the UI, REST, the CLI, agentic AI, and MCP. How a decision resolves.
Simba, our AI agent, now builds and manages every artifact. It creates and modifies all artifact types, integrates with Composer, our visual flow builder, and standard flows, and is tenant-aware from the CLI.
Parallel extraction. ETL flows can discover a table's primary key and split it into balanced ranges that extract in parallel, with no Partition SQL and no tokenized source query. Separately, a CDC snapshot of one large keyed table splits into non-overlapping key ranges handled by several threads at once.
Context-aware autocomplete arrived in the code editors, driven by the language, the cursor position, available fields, declared symbols, the functions and macros you are authorized to call, and the schema of the live database connection. Why it does not guess.
Also in this release: an AI usage and cost dashboard with per-user and per-tenant breakdown, Simba answering incoming support requests, a QVD connector for Qlik files, Databricks bulk-load throughput options, and SQL Server CDC reading the change tables directly.