Auditability & Observability

The Generative Studio X (GSX) platform provides comprehensive auditability and observability to support governance, compliance, and operational oversight. All AI agent actions, human interventions, and system events are captured in detailed, immutable logs with configurable access controls and retention policies. These capabilities enable organizations to monitor agent performance, enforce policies, and maintain a verifiable record of interactions for internal review, regulatory audits, and risk management.

Extensive Platform Logging

The platform is built on an event model, meaning it logs a vast amount of data from every conversation. Including, but not limited to, metrics such as:

Custom tags Time Location Context Transcript history Feedback User information (total users, new users, user engagement) Sentiment Session length (including response times) Conversation outcomes (completed, uncompleted, fulfilled, unfulfilled, escalated) Fallback responses Missed intents Repeat engagement Top messages and paths taken by users Time-based, user-based, and topic-based pattern recognition Hundreds of other configurable data points

GSX allows organizations to enrich logs with custom metadata and user-defined events. This enables precise tracking of agent actions and interactions in alignment with internal compliance and operational requirements.

The platform provides configurable reporting and dashboards that summarize agent and user activity. Metrics are available for operational review, compliance audits, and performance monitoring, with options to export data for further analysis.

GSX supports real-time observation of agent workflows, including alerts based on configured thresholds. Automated notifications can trigger human review or intervention when defined conditions are met.

Detailed logs capture all agent interactions, including human handovers, escalations, and performance metrics. User activity, session metadata, and context are recorded to ensure traceability and support compliance audits.

In-depth transcript analysis down to each message is available, showing the complete JSON from NLU and external integrations. The analytics solution supports user behavior analysis through funnels, Sankey diagrams, and heat maps.

The platform offers application monitoring and reporting via a support portal, with configurable alarm/alert message notifications and forwarding settings. The platform can integrate with external monitoring systems like New Relic, providing heartbeat monitoring and full observability. The platform allows for monitoring and alerting on custom performance and operation metrics for agents, including both technical (e.g., latency) and UX-related measures (e.g., abandonment rates, missed intents).

All actions, configuration changes, and data access events are logged with versioning and role-based access control. Audit trails support regulatory review and demonstrate enforcement of enterprise governance policies.

The platform can track where and why an agent conversation was handed off to a human agent. It also logs detailed agent performance metrics. Since AI agents can remain in the conversation even when a human agent is added, its ongoing monitoring capabilities can indirectly track agent behavior for compliance and training purposes.

The platform can capture the user's IP address and referring URL information. It can also track other user attributes and technical details if desired.