GoodData Cloud is rolling out a broad set of enhancements across visualization, data connectivity, automation, and Artificial Intelligence features. Geo analytics gains more control via a new generation of geo charts that are now generally available, including standalone geo area charts, support for multiple layers, and standard interactions such as drilling and cross-filtering. Editors can now disable panning and zooming and define default viewports for specific regions, while administrators can upload custom geo collections in Settings > Appearance & behaviour > Geo collection to model company territories or other custom polygons. These enhancements build on earlier experimental releases that introduced multi-layer geocharts, choropleth visualizations, and drilling support, all based on a modernized map engine that may introduce small visual differences but requires no user action.
Analytical Designer and dashboards receive several usability upgrades centered on filtering and table analytics. Attribute filters in Analytical Designer now support contains, starts with, ends with and their negative variants, and allow manual value entry for IDs or codes that are not preloaded in the value list, improving work with large text attributes. Metric value filters now support multiple numeric conditions combined using OR, allowing patterns such as values between 200 and 300 OR between 500 and 600, and they can be applied even to metrics not placed in buckets with configurable evaluation granularity. Date filters in dashboards and Analytical Designer can include or exclusively target empty or missing date values, helping highlight incomplete items, and they support custom time granularities and to-date options such as Year-to-Date, Quarter-to-Date, Month-to-Date, and Week-to-Date. Dashboards also benefit from Filter Grouping with collapsible sections, Saved Views, options to apply all filters at once, disabling cross-filtering per widget, and URL-based passing and sharing of filters so that specific dashboard states can be preserved and embedded.
Export and automation capabilities expand significantly. Scheduled exports now support dashboards as slideshow presentations in PDF or PPT, dashboard exports to XLS, raw tabular data delivery, and PNG exports of individual visualizations, complementing existing CSV/XLSX and new slide deck exports in PDF or PPTX. Pivot tables support paging, text wrapping, multi-column sorting, grand total row positioning, and direct PDF exports, while dashboards can be exported to Excel where each visualization appears on its own worksheet. Users can also export raw data in CSV form for large computations that exceed visualization limits, export widgets as PNG images, and display charts as tables via a Show as table toggle for more accessible tabular views. Centralized automations management provides organization or workspace level visibility into scheduled exports and alerts, including bulk operations, and alerts now feature customizable names, richer filter handling, per-dashboard and per-alert evaluation frequency, and support for notifications via email, webhooks, external recipients, and an in-app notification panel.
Artificial Intelligence capabilities are becoming more structured and governed. The Artificial Intelligence Assistant is generally available as a conversational interface that works through the semantic model, supporting anomaly detection on line charts, attribute-based filtering from approximate natural language values such as interpreting computers as Computers and Accessories, ranking queries like top 5 products by revenue, smart search integration, clarification questions when metrics or dimensions are ambiguous, and in-chat smart search widgets. The platform introduces Artificial Intelligence Memory, now generally available, letting teams define instructions, abbreviations, and synonyms so the assistant better understands internal terminology, and it can be tuned at the workspace level for consistent answers. Control over Artificial Intelligence visibility enables hiding technical attributes, labels, facts, metrics, and visualizations from Artificial Intelligence search and suggestions via an isHidden flag, while the Semantic Quality Agent scans titles and descriptions for identical or semantically similar text and unknown abbreviations, surfaces issues in the Analytics Catalog, and warns builders in the Artificial Intelligence Assistant when metadata quality could degrade responses.
Several features strengthen data modeling, localization, and performance. Data Localization lets dashboards, Analytical Designer, and Artificial Intelligence Assistant responses show localized attribute values using secondary labels without changing logical definitions, keeping filters consistent across languages and ensuring exported reports reflect user language preferences. Fiscal calendars are now generally available, allowing custom fiscal year start months, and date filter improvements include support for fiscal comparisons. Data Source Routing enables routing queries to different compute clusters that share storage and the Logical Data Model, improving concurrency, isolating heavy analytical workloads from standard dashboards, and enabling cost-optimized compute strategies. Aggregate Awareness is fully released, automatically selecting pre-aggregated or detailed datasets for faster queries and consistent drills. Debug metadata is now embedded into SQL queries so teams can map queries back to specific users, workspaces, dashboards, or visualizations, and resource utilization is optimized by canceling obsolete executions when users rapidly change filters or configuration.
GoodData continues to expand supported data sources and deprecate underused or externally constrained ones. CrateDB has moved from beta to full support as a data source for real-time and IoT analytics, and new native connectors were added for Amazon Athena, StarRocks, MotherDuck, ClickHouse, and FlexConnect, which allows integrating arbitrary computations and external data such as streaming feeds, benchmarks, anonymization logic, or machine learning predictions using Flight RPC. MongoDB initially reached general availability as a data source, but in line with MongoDB’s deprecation of the MongoDB BI Connector, the MongoDB connector in GoodData will be deprecated by the end of September 2026, while Apache Drill and Greenplum data sources will be removed by the end of March 2026. Snowflake data sources are transitioning to more secure authentication with key-pair credentials in response to Snowflake’s move away from single-factor password authentication, with legacy user types and basic authentication scheduled to stop working in November 2025.
On the developer and platform side, React SDK 11 introduces support for React 19 and discontinues support for React 16 and 17, placing React SDK 10 into a six-month support-only phase before end-of-life. Customers with custom dashboard plugins must update plugins to support React 18 and React 19 and ensure compatibility with dashboard upgrades, with multiple reminders to update plugins by specific dates such as February 22, 2026. A new generation of visualization features, including Dashboard Tabs, Column and Widget Containers, a Visualization Switcher, more flexible minimum widget sizes, grand total positioning, and rich text enhancements with references to live metrics and attributes, improves layout and content authoring. Entity APIs now accept application/json in addition to application/vnd.gooddata.api+json, rate limiting is being introduced with 429 responses for excessive calls, and a self-service audit log gives SecOps teams detailed visibility into user actions. The GoodData MCP Server, released as an experimental feature and later expanded, exposes analytics capabilities over the Model Context Protocol so Artificial Intelligence clients like Cursor or MCP-enabled chat tools can manage alerts, browse metadata, and generate visualizations using natural language, with updated toolsets for listing and managing objects and new alert management operations.
Security, identity, and governance workflows receive multiple refinements. Federated Identity Management now supports embedded contexts, multiple SAML providers, and Azure Identity Management, while identity provider configuration has been redesigned so organizations link to an identityProvider relationship instead of embedding OIDC attributes directly; legacy OIDC fields in the organization entity are deprecated and scheduled for removal, and an upcoming breaking change removes them entirely. Customizable JIT provisioning is fully supported, allowing organizations to define group mappings and scopes even when identity providers have limitations, and works alongside existing JIT implementations except in specific federated identity scenarios. PostgreSQL data source configuration now enforces ssl or sslmode parameters, defaulting existing sources to sslmode=prefer while recommending sslmode=verify-full, and a breaking change in the Python SDK introduced the USE_AI_ASSISTANT permission, requiring users of declarative permission APIs to upgrade the SDK to version 1.42.0 or later. Collectively, these changes modernize authentication paths, reinforce encryption in transit, and ensure Artificial Intelligence capabilities respect updated permission models while remaining backward compatible for most deployments.
