---
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


---

# Troubleshoot AI agents

# Troubleshoot AI agents {#ariaid-title1}

Release version: Australia  
Updated October 16, 2026  
![](https://www.servicenow.com/docs/portal-asset/ico-clock) 3 minutes to read
Summarize  
![AI sparkle icon](https://servicenow.com/docs/portal-asset/ai-sparkle-icon) Summarized using AI  
This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.  

## Summary of Troubleshoot AI agents

This guide assists ServiceNow customers in diagnosing and resolving issues related to AI agents within AI Agent Studio.
It emphasizes the importance of correct role assignments, testing procedures, and log analysis to ensure AI agents and agentic workflows function properly.
Understanding and managing user and data roles is critical to prevent permission-related failures or silent errors.
Show full answer Show less  

## Role Assignments and Access Control

* AI agents require specific user and data roles to access necessary tables and tools; these roles come predefined but can be customized by the customer.
* Altering default roles without proper understanding can disrupt agent functionality.
* Key AI Platform roles include issue readiness, domain separation, and access analyzer roles, each controlling distinct permissions.
* Customers can view assigned roles for AI agents and workflows via AI Agent Studio under the Access rules sections.

## Testing and Diagnosing AI Agent Failures

* Test all AI agents and agentic workflows individually to verify correct role assignments and resource access.
* Check that roles have appropriate resource access records and are not overridden by role filtering.
* Important diagnostic tables include:
  * **Generative AI Log \[sysgenerativeailog\]**: Central log capturing execution details, errors, and state transitions to identify failures.
  * **Generative AI Configurations \[sysgenerativeaiconfig\]**: Validates agent configuration before testing.
  * **AI Agent memory \[snaiamemory\]**: Useful for troubleshooting memory-related issues.
  * **AI Agent \[snaiaagent\]**: Helps verify agent definitions during failures.
* Silent failures returning no results often signal permission issues; ensure testing includes all underlying AI agents, not just workflows.

## Using Logs for Troubleshooting

The **Generative AI Log \[sysgenerativeailog\]** records all AI activity and is essential for diagnosing failures such as permission denials, endpoint unavailability, token limits, and performance issues. Key log columns to review include:

* **userrole**: Roles assigned to the agent compared against required roles.
* **errormessage**: Indicates permission or request issues.
* **permissiondenied**: Flags permission failures.
* **tablename** and **operationtype**: Identify the target table and operation causing the failure.
* **action**: Specifies failed steps such as create, update, or query operations.
* **executionlog**: Provides detailed context and reasoning before failure.
* **aclrule**: Shows the specific access control rule that denied access.

## Additional Resources

* Guides on configuring granular admin roles and implementing access control for AI Agent Studio.
* Tools for testing access and manually testing agentic AI assets.
* Comprehensive AI agents library and knowledge base for advanced scenarios and troubleshooting.
* Documentation covering security enhancements, common solutions, and upgrade considerations for AI agents.  
Use this troubleshooting guide to resolve issues with AI agents' roles, testing procedures, silent failure diagnosis and reference resources.

## Required role assignments {#troubleshoot-aia-new__required_roles}

Role assignments control access to AI Agent Studio and determine what tables and tools an agent or an agentic workflow can access.  
Each agent comes with predefined user and data roles by default that can be modified by the customer. It is important to understand the difference between these roles, as an AI agent may not function correctly if it lacks the appropriate user or data role assigned. For more information about the default AI agent roles, see the [ServiceNow AI agents library](https://www.servicenow.com/docs/jgwT0FAd6GNDiZr8mEkFvQ "All ServiceNow AI agents are listed. Depending on your license, not all AI agents may appear in your instance.").  
Note:  
If the customer alters the default user or data roles, it may lead to issues with the agent's functionality and execution.

Some of the individual ServiceNow AI Platform AI agent roles include:
{#troubleshoot-aia-new__table_sxl_s14_jkc__entry__2}

| AI agent | Role or Condition |
|-|-|
| Issue readiness | sn_uxc_gen_ai.platform_ai_issue_readiness |
| Request status | LoggedIn=true\^ORAllowUnauthRolelessAcl=true |
| Domain separation | domain_admin |
| Access analyzer agent | access_analyzer_admin OR user_admin |
[Table 1. ServiceNow AI Platform AI agent roles]

{#troubleshoot-aia-new__table_sxl_s14_jkc}

To identify the roles currently assigned to:

* AI agents: Navigate to AllAI Agent StudioAgentic solutions.
* Select an AI agent, and navigate to the Access rules section.

  For more information, see [Define access rules for an AI agent](https://www.servicenow.com/docs/i0oknNenWsCjujmAxvRMrw "Define security controls for an AI agent to determine which users can access it and what permissions they have.").
* Agentic workflows: Navigate to AllAI Agent StudioAgentic solutions.
* Select an agentic workflow, and navigate to the Access rules section.

  For more information, see [Define access rules for an agentic workflow](https://www.servicenow.com/docs/kgaLaIpme8WWfTULPnDFeQ "Define security controls for an agentic workflow to determine which users can access it and what permissions they have.").
{#troubleshoot-aia-new__ul_kfm_zwc_hkc}

## Diagnose AI agents' failures {#troubleshoot-aia-new__section_prg_1yz_fkc}

Test agents and workflows

:   Test all your AI agents and agentic workflows to make sure they have the correct access. When an AI agent role does not have required resource access, verify that the agent is assigned the correct role. Verify the role has resource access records for all required tables and confirm that role filtering does not override the resource access.Supporting tables for diagnosing AI agent's failures:

    * generative AI Log \[sys_generative_ai_log\] table: The primary resource for diagnosing AI agent failures is the generative AI Log \[sys_generative_ai_log\] table. The table logs agentic execution details, errors, and state transitions that help diagnose role-related failures, permission and workflow issues.
    * generative AI Configurations \[sys_generative_ai_config\] table: Contains configuration records for AI agents; useful for validating that agent setups are correct before testing failures.
    * AI Agent memory \[sn_aia_memory\] table: Contains the agent memory state and helps test memory-related failures or edge cases.
    * AI Agent \[sn_aia_agent\] table: Verify agent definitions during failure scenarios.

    {#troubleshoot-aia-new__ul_qd2_xc4_jkc}  
    Important:  
    Silent failures that return no values may indicate an underlying access error. You might have access to the agentic workflow, but the AI agent within the workflow may lack necessary permissions. Always test all underlying AI agents individually, not just the workflow, to catch permission gaps that your own access level might mask.

    To manually test an agentic AI asset, see [Manually test an agentic AI asset](https://www.servicenow.com/docs/SyWGZyDlSJheAEuP2QA5ig "Manually test an AI agent or agentic workflow to verify it functions as defined and achieves the desired objectives.").

Review logs
:   Use these columns in the generative AI Log \[sys_generative_ai_log\] table to identify root causes of AI agent and agentic workflow failures. The table records all generative AI activity within your ServiceNow instance, including AI agent invocations, model responses, performance metrics, and errors.  
    Common issues include:

    * Permission and ACL failures
    * Model endpoint unavailability
    * Token limit exhaustion
    * Skill or action failures
    * Performance degradation

    {#troubleshoot-aia-new__table_wdh_mz4_jkc__entry__2}

    | Column name | Troubleshooting use |
    |-|-|
    | user_role | Shows what role the agent has; compare to what's required |
    | error_message | Shows whether the error is a permission problem or a wording mismatch in the request |
    | permission_denied | Boolean flag to quickly filter permission failures |
    | table_name | Target table the agent was accessing when it failed |
    | operation_type | Type of operation attempted: `read`, `write`, `execute`, `query` |
    | action | Shows exactly which step failed, for example "update_incident," "create_change," or "query_table" |
    | execution_log | Full context; shows agent's reasoning before failure |
    | acl_rule | References the specific ACL that denied access |
    [Table 2. Columns from generative AI Log \[sys_generative_ai_log\] table used for troubleshooting AI agents]

    {#troubleshoot-aia-new__table_wdh_mz4_jkc}

## Additional resources {#troubleshoot-aia-new__section_fvg_qzz_fkc}

Use the following documentation resources for more information:

* Granular admin roles: [Configuring user access and data permissions for agentic workflows](https://www.servicenow.com/docs/access?context=cmpro-conf-users-agentic-wf&version=australia&pubname=australia-employee-service-management&ft:locale=en-US)
* Access control: [Implement access control in AI Agent Studio](https://www.servicenow.com/docs/4632C_BpTXjste_glQl_fg "Implement security in AI Agent Studio through access control lists (ACLs), user identities, and role filtering.").
* Access Analyzer: [Test access to an AI asset](https://www.servicenow.com/docs/qK1MtoBhHIdK_xctfMMfFQ "Verify whether a user has access to an agentic AI asset (an AI agent or an agentic workflow).").
* Manually testing user access to an agentic AI asset: [Manually test an agentic AI asset](https://www.servicenow.com/docs/SyWGZyDlSJheAEuP2QA5ig "Manually test an AI agent or agentic workflow to verify it functions as defined and achieves the desired objectives.").
* Agent library: [ServiceNow AI agents library](https://www.servicenow.com/docs/jgwT0FAd6GNDiZr8mEkFvQ "All ServiceNow AI agents are listed. Depending on your license, not all AI agents may appear in your instance.").
* Knowledge base: [AI agent advanced scenarios and runtime issues playbook](https://support.servicenow.com/kb?id=kb_article_view&sysparm_article=KB3042061).
* [AI Agents FAQ and Troubleshooting](https://www.servicenow.com/community/servicenow-otto-articles/ai-agents-faq-and-troubleshooting/ta-p/3200454).
* [Access control security enhancements for AI agents and Skill Kit](https://www.servicenow.com/community/servicenow-otto-articles/latest-access-control-security-enhancements-for-ai-agents-and/ta-p/3374036).
* [Common solutions for ServiceNow AI Agent, MCP Server, and AI Agent Fabric](https://support.servicenow.com/kb?id=kb_article_view&sysparm_article=KB2076278).
* [Accessing AI Agent Studio after upgrading the ServiceNow Otto AI Agents plugin](https://www.servicenow.com/community/servicenow-otto-for-creator/access-denied-to-ai-agent-studio-after-upgrading-sn-aia-plugin/td-p/3453111).
{#troubleshoot-aia-new__ul_w1r_wzz_fkc}

*[\>]: and then