AI asset data model attributes

  • Release version: Zurich
  • Updated July 31, 2025
  • 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 AI asset data model attributes

    This content details the additional attributes for AI asset data models within ServiceNow's Zurich release, focusing on product models, digital assets, and their key metadata. It helps ServiceNow customers understand how to structure and manage AI-related data such as models, datasets, prompts, and systems to support AI functionalities like response generation, training, evaluation, and deployment.

    Show full answer Show less

    AI Product Models

    • AI model product model: Contains metadata about AI models used autonomously by AI systems. Key attributes include model parameters, supported languages, model size, deployment guidelines, source information, training procedures (covering various machine learning techniques), and context window size (input token capacity).
    • AI dataset product model: Describes datasets used for training/testing AI models. It specifies data types (text, image, video, table), dataset sources, and acceptable usage according to licensing or contracts.
    • AI prompt product model: Captures information about instructions (prompts) given to AI models, including documentation on requirements and design.
    • AI system product model: Details software systems providing AI/ML capabilities to generate outputs such as decisions or recommendations, along with associated documentation.

    Digital Assets

    • Information asset: Defines data classification levels (e.g., public, confidential) following organizational policies.
    • AI digital asset: References ServiceNow Now Assist records and tables, linking AI-related records within the platform.
    • AI system digital asset: Links to associated AI models, evaluation datasets, and evaluation metric reports.
    • AI model digital asset: Includes references to base models, model weights, infrastructure requirements, training and evaluation datasets, evaluation metrics, license details, and model cards—especially for internally developed models.
    • AI dataset digital asset: Contains lineage information (base datasets), dataset cards describing data quality and distribution, and license details.
    • AI prompt digital asset: Contains prompt details and links back to the AI model for which the prompt is created.

    Practical Benefits for ServiceNow Customers

    By leveraging these comprehensive AI asset data model attributes, ServiceNow customers can:

    • Maintain detailed and standardized metadata for AI models, datasets, prompts, and systems within their CMDB.
    • Track AI model lineage, training, evaluation, and deployment details to ensure governance and compliance.
    • Facilitate integration and management of AI capabilities such as automated response generation and decision support.
    • Document infrastructure requirements and licensing information to support operational readiness and legal compliance.
    • Use these attributes to improve AI asset discoverability, auditing, and lifecycle management within the ServiceNow platform.

    Additional attributes for the AI asset data model.

    Attributes

    AI model product model: Product Information for the AI model that is used by the AI system to generate responses without human intervention (cmdb_ai_model_product_model).

    Table 1. AI model product model
    Attribute Description
    Model parameters info Number of parameters for the model.
    Supported languages Languages supported.
    Model size Size of the model in MB. Mostly applicable for models developed and deployed within the organization.
    Deployment guidelines Instructions applicable for models developed and deployed within the organization.
    Source Links or details of source of the model sources example: Hugging face, Microsoft, and so on.
    Training procedure Types of training
    • Decision Trees
    • Deep Neural Networks
    • Linear Regression
    • Logistic Regression
    • Random Forest
    • Supervised Learning
    • Unsupervised Learning
    • Reinforcement Learning
    • Transfer Learning
    • Semi-Supervised Learning
    • Instruction Finetuning
    • Supervised Finetuning
    Context window Size of input sequences that the model can handle (number of tokens).

    AI dataset product model: Product Information for the collection of data that is used to train and test AI models (cmdb_ai_dataset_product_model).

    Table 2. AI dataset product model
    Attribute Description
    Data type Describes data, example: Text, Image, Video, and Table
    Source Links or details of source of the dataset sources, example: Customer, Wikipedia, Hugging face, Crowd sourced, and so on.
    Acceptable usage Acceptable usage of the data according to license / contract example: Training, and Evaluation.

    AI prompt product model: Product Information for instructions given to AI models to get a response for AI system (cmdb_ai_dataset_product_model).

    Table 3. AI prompt product model
    Attribute Description
    Documentation Links and information about requirements, design, and related information.

    AI system product model: Product Information for software that provides ML / AI capability to generate outputs, such as decisions, recommendations, content, or predictions (cmdb_ai_system_product_model).

    Table 4. AI system product model
    Attribute Description
    Documentation Links and information about requirements, design, and related information.
    Table 5. Information asset
    Attribute Description
    Data classification Classification according to organization's data classification model, example: Public, confidential, and customer confidential.
    Table 6. AI digital asset
    Attribute Description
    ServiceNow® record reference Reference to Now Assist record.
    ServiceNow® table Now Assist table.
    Table 7. AI system digital asset
    Attribute Description
    AI models Reference to more than one associated models.
    Evaluation Dataset Reference to more than one associated datasets used for evaluation.
    Evaluation Metrics Report Details of evaluation results.
    Table 8. AI model digital asset
    Attribute Description
    Base model This AI model version was derived from an internal model developed within the organization.
    Model weights info Additional model information if available. Mostly applicable for models developed within the organization.
    Required infrastructure Documentation of infrastructure requirements for model deployment, primarily for models deployed within an organization. Example: Infrastructure stack and processing requirements.
    Training dataset Reference to one or more associated datasets used for training the model. These datasets are mostly applicable for models developed within the organization.
    Evaluation dataset Reference to one or more associated datasets used for evaluating the model. These datatsets are mostly applicable for models developed within the organization.
    Evaluation metrics report Links or details of evaluation results
    License details Link or detail to applicable licenses applied to the model.
    Model card Links to shareable model card (Internal and external model card).
    Table 9. AI dataset digital asset
    Attribute Description
    Base datasets

    This version of the AI dataset was derived from the previous version.
    Dataset card Information on number of records, distribution, and so on.

    Documentation for data quality and known risks and limitations.
    License details Link or detail to applicable licenses applied to the dataset, example: CommonCore, Apache 2.0,etc.
    Table 10. AI prompt digital asset
    Attribute Description
    Prompt information Details of the prompt.
    AI model

    Reference to the AI model for which the prompt is created.