Multi-instance Setup

  • Release version: Yokohama
  • Updated March 12, 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 Multi-instance Setup

    The Multi-instance Setup in AI Control Tower enables a production (prod) instance to centrally manage and communicate with multiple sub-production (sub-prod) instances. This setup uses the multi-instance framework to synchronize AI assets and rules, streamlining the review and management process across environments.

    Show full answer Show less

    Key Features

    • AI Asset Synchronization: Synchronizes AI assets such as systems, models, prompts, datasets, and from the September 2025 release, AI agents from sub-prod instances to the prod instance. This enables faster and unified asset review and control.
    • Version Compatibility: Starting May 2026, both prod and sub-prod instances must run the same AI Control Tower core version (minimum 6.2.4) to ensure the multi-instance framework functions correctly.
    • AI Inventory Management: The prod instance maintains the authoritative lifecycle state of assets, reflecting their status from a production standpoint. Asset states are not synchronized across environments; only assets activated in prod show as deployed.
    • Data Sharing Preference: When enabled, data sharing preferences set in prod are applied uniformly to all sub-prod instances. This preference is off by default.
    • Data Overflow Processing and Bursting Preference: These preferences can be enabled to apply prod-level settings to all sub-prod instances for consistent data handling. These settings are off by default.
    • Read-Only Preferences for Sub-prod: Once multi-instance management is enabled, sub-prod instances have read-only access to preferences configured at the prod level, ensuring centralized control.

    Practical Considerations for ServiceNow Customers

    • Ensure consistent AI Control Tower core version (6.2.4 or later) across prod and sub-prod instances before enabling multi-instance setup.
    • Use the multi-instance framework to synchronize AI assets and rules to maintain alignment between production and sub-production environments.
    • Understand that asset lifecycle states reflect production status only, preventing premature deployment indicators from sub-prod testing environments.
    • Leverage data sharing and data overflow preferences to maintain consistent data governance across all instances.
    • Note that sub-prod preferences are read-only under multi-instance management to centralize administration in prod.

    The multi-instance setup enables a prod (manager) instance to control, manage, and communicate with multiple sub-prod (managed) instances for AI Control Tower.

    AI asset Synchronization

    Multi-instance setup uses the multi-instance framework, which helps the user to synchronize assets from sub-prod instances to prod instances for a faster review process.

    Multi-instance setup synchronizes rules for the sub-prod instances from the prod instance.

    Note:
    Starting with the May 2026 release, confirm that both the prod and sub-prod instances are running the same AI Control Tower core version (6.2.4), which is the minimum supported version.

    If there’s any upgrade to version 6.2.4 in a sub-prod, then it’s advisable to upgrade the prod instance to 6.2.4 to confirm Multi-instance framework functions correctly.

    AI inventory information
    You can include the sub-prod instances that you want to synchronize with the prod instance. This synchronizes AI inventory information between the instances.

    When configured, the scheduled job starts synchronizing AI systems, AI models, prompts, and datasets. From the September (2025) release, the job has been enhanced to include synchronizing AI agents as well.

    Note:
    State of the assets while configuring Multi-instance management.

    The AI inventory in production reflects the true state of your assets like models, datasets, or skills from a production standpoint. Even if a model or dataset is active in a sub prod (lower) environment, it's still considered as under development from a prod perspective, since it's being tested and not yet live.

    For this reason, you don’t synchronize asset states across environments. An asset’s state changes to deployed only when the asset and its related records are activated in the production system.

    In summary, the state represents the overall lifecycle of the asset, not its local status in a specific environment.

    Data sharing preference
    You have the option to enable the data sharing preference, when it is enabled the preferences of the data sharing from the production will be applied to all sub-prod instances. By default, the data sharing preference is turned off.
    Data overflow processing and bursting preference
    You have the option to enable the data overflow processing and bursting preferences, when it is enabled the preferences of the data overflow and bursting from the production will be applied to all sub-prod instances. By default, data overflow processing and bursting is turned off.
    Note:
    All the preferences mentioned earlier for a sub-prod instance are available in read-only mode, when Multi-instance is configured and enabled.

    For information about configuring Multi-instance management for AI Control Tower, see Configure Multi-instance management for AI Control Tower

    For information about Data section, see Data sharing, Data overflow processing, and Security & privacy in AI Control Tower