AI capabilities for enhancing custom applications

  • Release version: Zurich
  • Updated July 31, 2025
  • 5 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 capabilities for enhancing custom applications

    Now Assist for App Engine offers powerful AI capabilities to enhance custom applications on the ServiceNow platform. These include Skills, AI Agents, and Agentic Workflows, each designed to automate tasks, generate insights, and improve user interactions within your custom apps.

    Show full answer Show less

    These capabilities leverage the ServiceNow AI Platform and large language models (LLMs) to deliver tailored AI-driven solutions, helping customers streamline processes and increase efficiency.

    Skills

    Skills are generative AI functions that perform specific tasks, such as summarizing records, based on input data and LLM prompts. Key components include:

    • Input: Data analyzed, typically records from tables.
    • Prompt: Instructions guiding the LLM on task execution and output style.
    • Output: Generated responses, which can be text, structured data, or formatted content.
    • Activation methods: Ways to trigger skills, including conversational interfaces, UI buttons, or Virtual Agent interactions.

    There are two types of skills:

    • Now Assist Platform skills: Preconfigured for common use cases but require significant customization to fit custom apps.
    • Custom skills: Built using the Now Assist Skill Kit, designed from the start to integrate into custom applications with defined inputs and outputs.

    Additionally, the custom app record summarization skill (available from version 28.2.4) allows AI-generated summaries of records in custom tables, configurable for specific fields and accessible via UI or conversation.

    AI Agents

    AI Agents are generative AI entities with defined roles, instructions, tools, and triggers that perform complex tasks. Components include:

    • Role: Defines agent capabilities and actions.
    • Steps: Sequential instructions sent to the LLM to guide task execution.
    • Tools: Supplementary resources like skills and data sources.
    • Trigger and availability: Determines how and where agents are activated and accessed (e.g., Now Assist panel or Virtual Agent).

    Now Assist AI agents are prebuilt for common scenarios but require notable modification to suit custom apps. Alternatively, you can create custom AI agents tailored to your use case using AI Agent Studio for seamless integration.

    Agentic Workflows

    Agentic workflows coordinate one or more AI agents to achieve specific business outcomes, such as task classification or email processing. They are managed by:

    • AI Agent Orchestrator: Controls workflow execution and agent coordination.
    • AI Agent Communicator: Facilitates communication between agents and the orchestrator.

    Like AI agents, Now Assist provides preconfigured agentic workflows for standard use cases that require customization to fit custom applications. You can also create custom workflows designed specifically for your needs.

    Practical Considerations for ServiceNow Customers

    • Choosing the right capability: Evaluate whether Skills, AI Agents, or Agentic Workflows best match your custom app’s use case and complexity.
    • Customization: Prebuilt Platform skills, AI agents, and workflows often require duplication and significant reconfiguration for custom app integration.
    • Custom creation: Using the Now Assist Skill Kit and AI Agent Studio enables building AI capabilities designed from inception to work seamlessly within your custom applications.
    • Activation and governance: Some skills might need data steward approval before activation, ensuring compliance and appropriate use.

    Leveraging these AI capabilities can greatly enhance automation, user experience, and operational efficiency within your custom ServiceNow applications.

    Learn about the AI capabilities available with Now Assist for App Engine that you can use to enhance custom applications, including skills, AI agents, and agentic workflows.

    There are several types of AI capabilities that you can add to custom applications with Now Assist for App Engine:
    • Skills
    • AI agents
    • Agentic workflows

    The following sections describe what each capability is and how you can use it to enhance a custom application. To learn about which AI capability might be best for your custom application use case, see Choosing the right AI capability.

    Skills

    On the ServiceNow AI Platform, a skill is a generative AI capability that contains instruction on how to complete a specific task for a use case, such as generating a summary.

    There are several important components of a skill.
    • The input is the data that the skill analyzes to generate an output. Typically on the ServiceNow AI Platform, inputs are records in tables.
    • The prompt is the instruction to the underlying large language model (LLM) that tells the skill what to do with the input. Prompts define the task, tone, and structure of the output.
    • The output is the generated response from the LLM based on the prompt and input. Outputs can be plain text, structured data, or even formatted responses depending on how the skill is configured.
    Skills also contain activation methods, which define how the skill can be accessed on the ServiceNow AI Platform. Skills can be called conversationally in the Now Assist panel, through UI actions such as buttons, and through conversations with Virtual Agent. To learn more about activating skills, see Activate a skill.
    Note:
    Some skills might need to be reviewed and approved by a data steward before you can activate them.

    Now Assist Platform skills and custom skills

    With Now Assist for App Engine, you have access to several kinds of skills: Now Assist Platform skills and custom skills. While both are able to be used in custom applications, the following table provides more information about each skill type and the ease of using the skill in custom applications.
    Table 1. Differences between Now Assist Platform skills and custom skills
    Skill type Description Ease of use in custom applications More information
    Now Assist Platform skills Preconfigured skills that are designed for existing use cases,such as document extraction and dashboard and visualization export Must be duplicated and reconfigured significantly for use within custom apps Now Assist skills in the Platform workflow
    Custom skills Skills that you create custom for your use case with Now Assist Skill Kit Can be designed to work within custom apps during the creation process, when you define the skill inputs and outputs

    Now Assist for App Engine custom app record summarization skill

    Starting with version 28.2.4 of Now Assist for App Engine, you can also use the custom app record summarization skill, which is a template skill available with Now Assist for App Engine. The skill enables you to generate AI summaries of records within custom apps and tables. You can configure the skill to summarize the tables and fields that you need, and you can choose how your app users interact with the skill through in-product UI or conversation in the Now Assist panel. For more information about the custom app record summarization skill, see Custom app record summarization skill.

    AI agents

    An AI agent is a generative AI capability that contains a set of large language model (LLM) instructions with the tools to perform specific tasks. On the ServiceNow AI Platform, AI agents are often part of agentic workflows and can be orchestrated to perform complex tasks. More information about agentic workflows is included in the following section.

    There are several important components of an AI agent.
    • The AI Agent role defines the capabilities and responsibilities for the AI agent. Roles enable the AI agent to perform its required actions.
    • The list of steps are the instructions are sent to the large language model (LLM) that describe the necessary steps the AI agent follows while carrying out its role.
    • The tools are the additional resources, such as skills, and data sources that you can provide your AI agent to accomplish its role.
    • The trigger defines how the AI agent is activated.
    • The availability determines how your AI agent is displayed, either in the Now Assist panel or Virtual Agent.
    For more information about AI agents, see AI Agent Studio.

    Agentic workflows

    An agentic workflow contains a set of large language model (LLM) instructions with one or more AI agents that can execute an objective. Agentic workflows are tailored to achieve specific business outcomes, such as classifying tasks or processing emails for tasks.

    An agentic workflow is controlled by an AI Agent Orchestrator that coordinates the flow of work between AI agents. Additionally, an AI Agent Communicator helps to facilitate communication between the AI Agent Orchestrator and AI agents in the workflow.

    For more information about agentic workflows, see General guidelines for creating AI agents and agentic workflows.

    Now Assist AI agents, Platform agentic workflows, and custom AI agents and agentic workflows

    With Now Assist for App Engine, you enhance custom applications with Now Assist AI agents and Platform agentic workflows, in addition to custom AI agents and agentic workflows. Like with Now Assist skills, configuring Now Assist AI agents and Platform agentic workflows to fit custom applications often takes more time than just building custom AI agents and agentic workflows. The following table provides additional information about Now Assist AI agents, Platform agentic workflows, custom AI agents, and custom agentic workflows.
    Table 2. Differences between Now Assist AI agents and agentic workflows and custom AI agents and agentic workflows
    Ease of use within custom applications More information
    Now Assist AI agents and agentic workflows Preconfigured AI agents and agentic workflows that are designed for existing use cases, such as incident resolution Must be duplicated and modified significantly for use in custom applications
    Custom AI agents and agentic workflows AI agents that you create custom for your use case using AI Agent Studio Can be designed to work within custom apps during the creation process, when you define the AI agent and agentic workflow