Using Now Assist for App Engine

  • Release version: Australia
  • Updated March 12, 2026
  • 1 minute to read

    • Enhance custom applications with generative and agentic AI capabilities.

    With Now Assist for App Engine, you can enhance your custom applications with several AI capabilities: skills, AI agents, and agentic workflows. You can implement any number and combination of AI capabilities within your custom apps, depending on your app's workflow. To learn more about the types of AI capabilities that you can implement with Now Assist for App Engine, see AI capabilities for enhancing custom applications.

    Certain app workflows might be well suited for AI agents and agentic workflows. Other workflows might be best handled with skills. To learn about which AI capability might be best for your use case, see Choosing the right AI capability.

    Implementing skills

    The following list outlines the process for implementing a skill within a custom application:
    1. Create a skill or configure a Now Assist Platform skill for your app's use case.
    2. Create a prompt to choose what skill inputs to use and the type of tool.
    3. Test the prompt to verify that you’re seeing the expected prompt results before the skill is activated.
    4. Evaluate the prompt to assess the effectiveness of your skill prompts.
    5. Finalize and publish the skill to enable a Now Assist admin to activate it.
    6. Activate the skill, enabling you to trigger the skill within the UI.
    7. Use the skill in your custom application.
    Note:
    Some skills might need to be reviewed and approved by a data steward before you can activate them.

    Implementing AI agents and agentic workflows

    The following list outlines the process for implementing an AI agent and agentic workflow within a custom application:
    1. Create an AI agent or duplicate an AI agent and modify it for your app's use case.
    2. Add a tool to the AI agent to enable different functionalities and help your AI agent to achieve their objective.
    3. Test the AI agent to see that it functions the way that you defined it.
    4. Create an agentic workflow or modify an agentic workflow so that AI agents can coordinate to solve complex problems.
    5. Test the agentic workflow to analyze its performance while it runs the instructions that you defined.
    6. Evaluate the agentic workflow to monitor performance and evaluate it against different benchmarks.
    7. Use the AI agent or agentic workflow in your custom application.