Coginiti 26.2 delivers major platform updates for automation and AI workflows, makes the Semantic Layer generally available, improves scheduling across daylight saving transitions, expands Snowflake authentication options, and ships multiple UX improvements across Object Stores and the Results Grid. This release also updates all supported LLM providers and includes important security and stability fixes.
Note: The Coginiti MCP Server will be released independently next week.
Semantic Layer is Now Generally Available
The Coginiti Semantic Layer is now production-ready and fully supported. Building on the public preview introduced in 25.11, the Semantic Layer now includes full SQL transformation support in the execution engine, including CTAS, subqueries, custom JOINs, and function calls.
The REST API surface has been significantly enhanced. The unified /execute API now supports SQL directly without requiring a separate semantic execution endpoint, and the /metadata and /codegen APIs have been expanded to support richer query models and discovery workflows.
This release also includes a new ODBC driver for the Semantic Layer, enabling direct connectivity from tools like Power BI and Excel. Teams can now access governed metrics and business logic from the tools they already use — without building custom integrations or duplicating definitions.
You can now standardize metrics and business logic once and expose them cleanly to BI tools, applications, and AI agents through a stable API surface — without semantic drift across tools.
Public Catalog Management API
A new public HTTP API enables programmatic management of SQL and CoginitiScript assets in the Analytics Catalog. You can now create, update, and delete folders, SQL scripts, and CoginitiScript documents through the API, with ETag-based optimistic concurrency control to prevent accidental overwrites.
The API supports asset type auto-detection, path-safe operations with validation, and full permissions enforcement. It is designed for automation tools and agent workflows, including n8n, AI agents, and CI pipelines.
Schedules Now Respect Local Wall-Clock Time Across DST
Schedules now run at consistent local times year-round using IANA timezones. Schedules created with timezones like America/New_York continue to run at the same local time through DST transitions. Existing schedules defined using UTC offsets are automatically migrated to proper IANA timezones. Your daily 9:00 AM job now runs at 9:00 AM — all year — without manual fixes twice a year.
AI and LLM Integrations
All supported LLM providers have been updated to the latest model versions.
New LLM Provider: Ollama
You can now use Ollama as an LLM provider for local and self-hosted model workflows. This enables offline and air-gapped AI workflows, supports regulated and sovereign environments, and reduces dependency on external SaaS model providers.
Snowflake Authentication Enhancements
Coginiti now supports External OAuth Identity Providers and key-pair authentication for Snowflake connections. These additions improve enterprise security posture, support regulated environments and SSO requirements, and enable service-to-service authentication without interactive logins.
Object Store UX Improvements
Working with Object Stores is now faster and more intuitive. You can drag and drop files from Object Stores directly to Local Files, filter buckets to show only frequently used buckets, and search buckets in the Object Stores filtering window. These changes reduce friction when moving data between object storage and local workflows, especially when working across multiple buckets and environments.
Results Grid Improvements
The Results Grid now offers more spreadsheet-like ergonomics for exploratory analysis and data shaping. You can autosize columns to fit cell contents, drag rows to manipulate result ordering, and delete columns directly from the grid — without round-tripping back to SQL for basic transformations.
Security and Stability
This release includes security patches and dependency updates, bug fixes across catalog operations, scheduling, and execution services, performance improvements in semantic query execution paths, and improved validation and error handling for public APIs.
What’s Coming Next
The Coginiti MCP Server is releasing independently next week, enabling deeper agent integration patterns and secure, structured AI interaction with Coginiti APIs and the Semantic Layer.
