AI Hub

AI Hub is an organization-level configuration section where administrators can configure and govern AI capabilities in GoodData. From AI Hub, you can manage AI agents, organization-level AI Knowledge, LLM providers, AI observability, and other AI settings.

AI agents control how the AI Assistant behaves, which skills it can use, whether it can access AI Knowledge, and which user groups receive that experience. Changes to agent configurations take effect immediately.

AI Hub Overview

The AI Hub page contains capability tiles, AI observability metrics, and AI settings. When AI Hub is enabled, the AI settings that were previously available in the main organization Settings section are managed from AI Hub instead.

The capability tiles include:

• Agents: opens the Agents page, where you can create, configure, and manage AI Assistant agent configurations.
• AI Knowledge: opens the organization-level AI Knowledge page, where you can upload and manage documents that the AI Assistant can use as trusted context.
• MCP: opens the documentation for connecting a custom MCP client to the GoodData MCP Server. MCP configuration is not managed in the GoodData UI.

The AI Observability section gives organization administrators a quick view of AI adoption and usage without leaving AI Hub.

The settings section includes:

• LLM providers: lets you add and manage providers, select models, and set defaults for the organization.
• Share data with AI: lets you control the organization-level data sharing setting for AI features that support data sharing.
• Query limits: lets you set organization-level limits for how many AI queries users can make within a defined time window.

AI Knowledge

When AI Knowledge is available for your organization, click Manage on the AI Knowledge tile to open the organization-level AI Knowledge page.

AI Observability

AI Observability helps organization administrators understand whether AI features are being used and how adoption changes over time. The AI Hub landing page includes three aggregate observability tiles so administrators can monitor AI usage directly from the same place where they configure agents, AI Knowledge, model providers, and data-sharing settings.

The observability tiles include:

• AI Queries: the number of successful AI actions performed in the organization during the current calendar month. One successful click or request counts as one AI query.
• AI Workspaces: the number of workspaces that had at least one AI action during the current calendar month, shown out of all workspaces.
• AI Users: the number of users who performed at least one AI action during the current calendar month, shown out of all users.

The AI observability tiles show cumulative running totals for the current calendar month. Each value starts at 0 on the first day of the month and increases during the month as users perform AI actions. The value is not a daily rate and not a rolling time window.

When available, the tile can also show a comparison with the previous calendar month. Interpret this comparison carefully early in the month. For example, a value on the third day of the current month is being compared with the full previous month, so a lower value at that point does not necessarily indicate a usage drop. During the first month when previous-month data is not available, the comparison is hidden.

These metrics are intended as a quick health overview. They help administrators identify whether AI rollout is gaining traction and whether users are engaging with AI features across workspaces.

AI observability in AI Hub shows aggregate usage metrics. It does not expose individual conversation content from the AI Assistant.

What Counts as an AI Query

The AI Queries tile counts successful AI actions, including:

• a message sent to an AI agent, regardless of whether it comes from the UI, an HTTP request, or another agent call
• using AI to generate content, such as a title or description
• using AI to perform analysis, such as change analysis or a quality check

The following actions do not count as AI queries:

• uploading knowledge documents
• failed AI requests
• browsing, opening, or reading past AI results without running a new AI action

Organization-Level AI Settings

The LLM providers and Share data with AI settings were moved from the workspace configuration to the AI Hub and show organization-level values. These values are used as defaults for AI features unless a workspace has its own override.

Workspace-level settings continue to take priority over organization-level settings for AI features used in that workspace.

Configure AI Assistant Agents

In AI Hub, you can create named agent configurations for the AI Assistant. Each agent configuration defines a specific assistant experience for a set of users.

An agent configuration includes these settings:

• Personality: a custom instruction that shapes how the assistant communicates
• Skills: the set of platform skills the assistant is allowed to use
• AI Knowledge: whether the assistant can access AI Knowledge
• Access: which user groups receive that assistant experience

This lets you create different assistant experiences for different groups of users without changing code or redeploying your solution. For example, you can create separate assistants for basic, advanced, or specialized use cases.

Create an Agent Configuration

1. Open AI Hub at the organization level.

2. In the Agents tile, click Manage.

3. Click Create agent.

4. Enter a name and, optionally, a description.

5. Configure the assistant:

   • choose which skills are enabled
   • add a personality instruction
   • turn AI Knowledge access on or off
   • assign the agent to one or more user groups, or enable it for all users

6. Save the configuration.

The new agent becomes active immediately after you create it.

Personality

Personality is a free-text instruction that influences how the assistant communicates. You can use it to adjust tone, style, or response behavior for a specific audience. For example, you might configure one assistant to provide short business-friendly summaries and another to provide more detailed analytical explanations.

Good personality instructions are specific, practical, and focused on communication style or decision rules.

Examples of useful instructions:

• Use concise business language and include the active filter context when explaining results.
• If the user's request is ambiguous, ask one clarification question before creating a visualization.
• For finance users, explain acronyms on first use and call out assumptions clearly.
• Prefer short summaries first, then offer more detail if the user asks for it.

Skills

You can control which platform skills an assistant is allowed to use. There are two modes:

• All enabled: the assistant can use all available platform skills
• Selected: the assistant can use only the skills you select

This helps you tailor the assistant to different user groups or use cases. For example, one assistant can have a limited skill set for general business users, while another can expose a broader set of capabilities for advanced users.

The exact list of available skills depends on the AI features enabled for your organization and workspace. Available skills may include:

• Alert creation: creates metric alerts with automatic notifications.
• Anomaly detection: finds anomalies, outliers, and unusual patterns in your data.
• Clustering: groups data points into clusters based on similarity.
• Dashboard summary: summarizes dashboard tabs or dashboard content.
• Data visualization: creates charts, tables, and visualizations from metrics and dimensions.
• Description: generates or improves business-friendly descriptions for analytics objects.
• Forecasting: forecasts future metric values based on historical data.
• Key driver analysis: interprets why a metric changed between two time periods.
• Knowledge search: searches the organization’s knowledge base for documents and information.
• Metric creation: creates MAQL metrics (measures, KPIs, calculated fields).
• Scheduled export: creates scheduled exports for recurring delivery.
• Visualization summary: summarizes ad-hoc or saved visualizations.
• What-if analysis: compares projected metric values under different business assumptions.

AI Knowledge

You can turn AI Knowledge access on or off for each agent configuration. When AI Knowledge is enabled, the assistant can use the document retrieval layer. When it is disabled, the assistant does not use that content source.

Enable AI Knowledge when the agent should answer using trusted documents, such as product documentation, internal guides, playbooks, or business definitions.

Disable AI Knowledge when:

• the agent should rely only on governed analytics metadata and data
• the assigned users should not receive document-based guidance
• the available documents are not curated for the target audience
• you want to limit the agent to a narrower, task-specific experience

Access by User Group

Agent configurations are assigned through user groups or by enabling access for all users. This allows administrators to manage assistant experiences at scale. For example, you can assign one assistant to an internal analyst group and another to a broader business user group. Individual user assignment is not supported in this scope.

Manage Existing Agents

AI Hub also lets you manage existing agent configurations. You can view existing agents, edit an agent, duplicate an agent, enable or disable an agent, and delete an agent. Changes take effect immediately after you save them.

Edit and Preview an Agent

When you edit an agent, you can update its name, description, personality, skills, AI Knowledge setting, or assigned user groups. In the agent detail, you can preview the agent and test its behavior, including unsaved changes, before you save the updated configuration.

Duplicate an Agent

Duplicating an agent creates a new configuration based on an existing one. This is useful when you want to create a similar assistant experience and adjust only a few settings. For example, you can duplicate a general-purpose assistant and then create a more specialized version for a specific user group. The duplicate is set to enabled by default.

Enable, Disable, or Delete an Agent

You can enable or disable agents to control which configurations are available to users, and you can delete configurations that you no longer need. A disabled agent is not available to assigned users until it is enabled again. Deleting an agent removes that configuration entirely.

How Agent Resolution Works

Users always see the AI Assistant. They do not see which agent configuration is active.

If multiple agent configurations match the same user, the most recently modified one is used. This lets administrators update assistant experiences centrally without requiring any action from end users.

Example

A common use case is to provide different assistant experiences for different user groups. For example:

• a general business group receives an assistant with a limited set of skills
• an analyst group receives an assistant with a broader skill set
• a documentation-focused group receives an assistant with AI Knowledge enabled

All of these configurations can be managed from AI Hub.