Build Agent limitations

  • Release version: Zurich
  • Updated June 15, 2026
  • 2 minutes to read
    • Summarize
    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 Build Agent limitations

    The Build Agent in ServiceNow enhances development workflows by automating code generation and application enhancements, but it has specific limitations that affect deployment capabilities and performance. Understanding these constraints helps customers plan deployments effectively and troubleshoot issues by complementing, rather than replacing, manual development steps on the ServiceNow AI Platform®.

    Show full answer Show less

    General Limitations

    • Human Review Required: AI-generated code must undergo thorough human review for access control (ACL) accuracy, security, platform standards, and governance compliance. Customers should implement peer reviews, static code analysis, and validation steps.
    • Partial Metadata and Feature Support: Build Agent does not yet support all app files or metadata types such as certain tables, user interfaces, or advanced features. Developers still need to use ServiceNow Studio builders for unsupported areas.
    • Security Model Scope: Security in apps built via Build Agent applies at the application and API levels by default. Fine-grained record- or field-level security requires additional agentic development steps.
    • Limited Cross-Product AI Integration: While Build Agent can generate data models and business rules, understanding where AI-generated app data is stored on the ServiceNow AI Platform is important for integration planning.

    Relationship with ServiceNow Fluent

    Build Agent capabilities align with the metadata types supported by the ServiceNow Fluent API. If a metadata type is unsupported by Fluent, Build Agent cannot update it unless customers provide precise platform-specific instructions. Reviewing the ServiceNow Fluent documentation helps clarify Build Agent’s upper functional limits.

    Regulated Environments Considerations

    • Build Agent v2 requires off-instance services that have not passed regulatory compliance for hosting environments such as GCC, NSC, and FedRAMP.
    • Customers in regulated environments must continue using Build Agent v1, which operates on-instance and is certified for compliance.
    • In these environments, development options include Developer Sandboxes, core update sets, Git workflows in ServiceNow Studio, Instance Scan, and ReleaseOps.

    Troubleshooting and Feedback

    If Build Agent cannot perform a specific task, customers should complete that task directly on the ServiceNow AI Platform and continue using Build Agent for other development activities. For troubleshooting guidance, refer to “Issues and solutions in Build Agent.” Customers can submit feedback and improvement ideas via the Idea portal on Now Support.

    Plan deployments and troubleshoot issues by learning about Build Agent constraints that affect deployment capabilities and performance.

    Troubleshooting

    If you encounter something that Build Agent can't currently do, complete that step directly on the ServiceNow AI Platform® and keep working with Build Agent for the rest of your development. Build Agent is designed to complement your workflow, not to replace it.

    For details on troubleshooting, see Issues and solutions in Build Agent.

    General limitations of Build Agent

    Be aware of the following limitations when using Build Agent:
    • AI-generated code requires human review for ACL correctness, security, platform conventions, and governance compliance. Put peer-review gates, static checks, and validation steps in place.
    • You still must work with builders in ServiceNow Studio for app files and metadata types not yet supported in Build Agent, such as certain tables, UIs, or advanced features.
      Note:
      Build Agent can work on existing applications. You can use Build Agent to enhance base workflows, such as creating business rules on existing tables.
    • Security in apps developed with Build Agent is at the application and API level, not at granular record and field level by default. However, you can use agentic development to build security onto the app by making requests. For examples, see Example prompts.
    • Limited support for cross-product AI integration.
    Build Agent can generate data models and business rules. You should understand where data from your AI-generated apps is stored on the ServiceNow AI Platform.

    Build Agent and ServiceNow Fluent

    To understand the upper limit of what Build Agent can do, review the ServiceNow Fluent documentation. If you're not familiar with the ServiceNow AI Platform, the ServiceNow Fluent documentation can help you determine what Build Agent can do. For more information, see the following topics:

    If ServiceNow Fluent does not support a metadata type, Build Agent cannot update it unless you provide precise platform-specific instructions.

    Regulated environments

    Build Agent and Test Agent depend on off-instance services that have not completed the security compliance review required for regulated hosting environments. As a result, Build Agent v2 (Australia Patch 0 and Zurich 8 and higher) is not available in GCC, NSC, or FedRAMP environments.

    Customers in regulated environments must remain on Build Agent v1, which runs on-platform and is certified for regulated use.

    In regulated environments where Build Agent v2 is unavailable, you can continue to use the following:

    • Developer Sandboxes
    • Core update set and Git workflows in ServiceNow Studio
    • Instance Scan
    • ReleaseOps

    Feedback on Build Agent

    To give feedback on Build Agent, see the Idea portal on Now Support.