--- sourceDocument: Australia Enable AI sourceDocumentLink: https://www.servicenow.com/docs/r/australia/intelligent-experiences Release : - australia ft:locale : - en-US ft:publication_title : - Australia Enable AI ft:clusterId : - platai bundleId : - platai workflow : - Platform --- # Activate evaluation scoring # Activate evaluation scoring for external AI systems {#ariaid-title1} Release version: Australia Updated July 13, 2026 ![](https://www.servicenow.com/docs/portal-asset/ico-clock) 3 minutes to read Assess the quality and safety of your external AI systems by connecting them to AI Control Tower and activating evaluation scoring. ## Before you begin Role required: sn_ai_governance.ai_steward ## About this task External AI systems require trace data to be scored by the evaluation engine. Unlike AI systems on the ServiceNow AI Platform, which are automatically instrumented, external AI systems must be connected to send trace data to AI Control Tower. There are two ways to establish this connection: * **SDK instrumentation**: Instrument your AI agent code using the Traceloop SDK and authenticate with an API key. Use this method for any AI agent framework that is not hosted on a supported cloud platform or monitoring service. * **Trace connection**: Connect to a supported cloud platform or monitoring service through your cloud credentials and a MID Server. Use this method when your AI agents already run on or send traces to a supported platform. No API key or code instrumentation is required. ## Procedure 1. Navigate to AllAI Control TowerHomeSettingsRules and templatesEvaluation. 2. On the AI evaluations sub-tab, select External AI systems. 3. Set the Activate toggle to on. 4. Connect your external AI system to send trace data to AI Control Tower using SDK instrumentation or a trace connection. | Connection method | Description | | SDK instrumentation | Use for any third-party AI agent framework (CrewAI, LangChain, AWS Agent Core, and others) not connected through a trace connection. Requires generating an API key and configuring the Traceloop SDK in your agent code. 1. Generate an API key to authenticate trace data requests from your external AI system. Note: Multiple API keys can exist, but only one is active at a time. 1. In the API key for external AI system evaluation section, select Create API key. The API Key record page opens on a new browser tab. 2. On the form, fill in the fields. {#mon-ai-monitor-external-ai-system__table_api_key_fields__entry__2} | Field | Description | |-|-| | Name | Name for the API key. | | Description | Purpose of the API key. | | User | User associated with this API key. Select a user with the `sn_ai_observe.ai_data_sender` role. Use a dedicated service account rather than a user with elevated permissions. This role is installed with the AI Control Tower Evaluations plugin and grants permission to send observability data without granting read or write access to AI Control Tower configuration. If the service account does not already have this role, assign it before creating the API key. | | Active | Whether the API key is active. | | Auth Scope | Scope that controls the API key's authority at runtime. | | Expiry | Date when the API key expires. Leave empty for no expiration. | [Table 1. API Key form fields] {#mon-ai-monitor-external-ai-system__table_api_key_fields} 3. Select Submit. 4. Copy the generated token and store it securely. The token is displayed only once. If you lose it, generate a new key, which deactivates the previous one. 5. Return to the AI Control Tower browser tab and refresh the page. The new API key appears in the API key for external AI system evaluation section. {#mon-ai-monitor-external-ai-system__ol_qj2_4cx_bjc} 2. Instrument your AI agent to send trace data to your ServiceNow instance. For instrumentation steps using the Traceloop SDK, see [Instrument a third-party AI agent with the Traceloop SDK for AI Control Tower Evaluations](https://support.servicenow.com/kb?id=kb_article_view&sysparm_article=KB3144255). {#mon-ai-monitor-external-ai-system__ol_pj2_4cx_bjc} | | Trace connection | Use when your AI agents run on a supported cloud platform or already send traces to a supported monitoring service. The ServiceNow AI Platform collects trace data through your cloud credentials and a MID Server. No API key or SDK instrumentation is required. See [Configuring trace connections](https://www.servicenow.com/docs/ryWU9mayy2W0VEuV6UVRnA "Collect trace data from AI agents running on supported cloud platforms and monitoring services to generate evaluation metrics, populate AI inventory, and support security monitoring in AI Control Tower."). | |-|-| {#mon-ai-monitor-external-ai-system__choicetable_connection_method} {#mon-ai-monitor-external-ai-system__steps_monitor_ext} ## Result Evaluation scoring is active for external AI systems. Scores appear on the monitoring Overview page as new sessions are evaluated. ## What to do next To add or remove global metrics and adjust their sample rate, see [Configure global metrics for external AI systems](https://www.servicenow.com/docs/nCY8JZLeIBAsDqQhk4Rfrg "Choose which metrics evaluate all your external AI systems by default, exclude an individual system from a specific metric, and adjust how often each metric runs."). **Related tasks** * [Configure global metrics for external AI systems](https://www.servicenow.com/docs/nCY8JZLeIBAsDqQhk4Rfrg "Choose which metrics evaluate all your external AI systems by default, exclude an individual system from a specific metric, and adjust how often each metric runs.") * [Configure asset-specific metrics for external AI systems](https://www.servicenow.com/docs/UWgY4DmrUn_VWbtGsLJI~g "Add or remove metrics for one or more external AI systems without changing the global metric configuration, and adjust metric sample rates.") * [Exclude external AI systems from a metric](https://www.servicenow.com/docs/htwTgh7pXSExNOqECMwdEA "Exclude one or more external AI systems from a specific metric, without changing that metric's configuration for every other system.") *[\>]: and then