AI Observability Workspace

AI Observability Workspace gives organization administrators a ready-made workspace for monitoring how AI features are used across their GoodData organization.

The workspace helps you track AI adoption, usage, reliability, and token consumption. You can see which workspaces and users are using AI, how usage changes over time, how many conversations are taking place, how different agents are used, and where errors, timeouts, or high token consumption occur.

The observability data is collected automatically as users interact with GoodData AI features. You do not need to add your own instrumentation.

It is exposed as a standard GoodData workspace, so the metrics and dashboards can be explored, filtered, and extended with the familiar analytics tooling.

Availability

AI Observability Workspace is a paid, entitlement-controlled feature. To enable it, contact GoodData Support or your GoodData account team. GoodData enables the necessary entitlement and provisions the required observability resources for your organization.

What GoodData Provisions

When AI Observability Workspace is enabled, GoodData provisions:

• a managed AI Observability workspace (ID: gd_obs) with a predefined logical data model, metrics, and dashboard
• a managed observability data source (ID: ai-observability) that exposes your organization’s AI usage data

The workspace and data source are scoped to your organization.

The workspace object and its logical data model are managed by GoodData. You can work with the analytical content in the workspace, including adjusting the bundled visualizations, building your own metrics and insights on top of the managed model, and arranging dashboards as you like.

Explore the AI Observability Dashboard

The bundled dashboard is organized into several tabs. The exact visualizations may evolve during the beta. Currently, the following tabs are available:

• The Overview tab provides a high-level view of AI activity and health across the organization.
• The Conversations tab focuses on chat engagement.
• The Agents tab breaks down AI activity by agent.
• The Token usage tab helps you understand AI consumption.

Build Custom AI Observability Dashboards

You can reuse the managed AI Observability data source in another workspace.

This is the recommended approach when you want to create custom metrics, visualizations, or dashboards without risking that they are reset when the managed workspace is updated.

The observability data source is read-only and exposes only the managed observability datasets.

The following restrictions apply:

• You cannot write data to the observability data source.
• Custom SQL datasets and arbitrary SQL queries are not supported.
• You cannot create custom table-backed datasets that point to arbitrary schemas or tables.
• Result caching is disabled so queries read current observability data.
• Normal connection and query limits apply.

Update the Managed Dashboard

The bundled dashboard does not update automatically when GoodData publishes a newer AI Observability layout. An organization administrator can reapply the latest managed layout by calling:

POST /api/v1/actions/organization/reloadObservabilityLayout

The request requires:

• the AI_OBSERVABILITY entitlement
• organization MANAGE permission

The action updates only the managed AI Observability Workspace.

Data and Privacy

The dashboards focus on usage counts, trends, sources, workspaces, users, and feedback. They do not provide full trace inspection or conversation-level drill-down.

Users see only AI usage data for their organization.

Relationship to AI Hub Observability Tiles

AI Hub can show a compact observability overview, such as AI queries, AI users, and AI workspaces. The AI Observability Workspace is the deeper analytics surface for users who need more detail, filtering, and exploration.

Use AI Hub for a quick status overview. Use the AI Observability Workspace for deeper analysis of AI usage.

Current Beta Scope

The beta release focuses on organization-level AI usage analytics.

It provides aggregate observability for AI adoption, conversations, agents, token consumption, reliability, and performance.

Interaction-level debugging of individual AI responses and AI-generated recommendations based on observability data are not part of this beta release.