---
sourceDocument: Australia Build or modify applications
sourceDocumentLink: https://servicenow-prod.fluidtopics.net/r/application-development

 Release :

    - australia

ft:locale :

    - en-US

ft:publication_title :

    - Australia Build or modify applications

ft:clusterId :

    - cadev

bundleId :

    - cadev

workflow :

    - Development, Data, and Analytics


---

# General guidelines for agentic development

# General guidelines for agentic development {#ariaid-title1}

Release version: Australia  
Updated March 31, 2026  
![](https://www.servicenow.com/docs/portal-asset/ico-clock) 1 minute to read  
When using agentic development tools to build applications and app components on the ServiceNow AI Platform, the quality of the output depends on the quality of your instructions. Use these guidelines to write effective instructions and improve the products of your conversations with AI tools.

A prompt or instruction is the primary input an AI tool uses to determine what to produce. Unlike a keyword search, generative and agentic AI treats your input as a task to complete. The quality of the output depends directly on the
quality of the input. Therefore, instructions that are clear and specific produce the most accurate, usable results. A well-constructed instruction includes a goal, context, expectations, and optionally a source. Use the following
guidelines to help you draft your instructions.

Goal
:   A goal describes what you want the AI to produce or do. In describing your goal, lead with an action verb and be specific. For example, <kbd class="ph userinput">Create a table for travel requests with fields for employee name, destination, and trip duration</kbd>.

Context
:   Context describes the purpose of the application and the types of users it serves. Context helps the AI to generate appropriate roles, permissions, and interfaces.

Expectations
:   Expectations describe how the application or output should look and sound. In your expectations prompts, specify tone, format, and structure.

Source
:   If there are specific information or resources the AI should use to generate your application or app component, describe those sources in your instructions. For example, direct the AI to specific tables or existing knowledge
    articles.

When working with Build Agent or ServiceNow Otto for Creator skills such as app generation, these guidelines apply across the entire conversation, rather than just a single prompt. For detailed guidance and examples, see [General guidelines for using app generation](https://servicenow-prod.fluidtopics.net/zo8fuhKg6AJr2aWT~39iMQ "To get the best results from app generation, provide clear and detailed descriptions of your application requirements during your conversation with ServiceNow Otto.").  
Important:  
Regardless of which AI tool you use, be sure to review AI-generated output before deploying the application or app component to a production instance. Verify that tables, fields, flows, scripts, and access controls match your
requirements, and test generated logic to confirm that it behaves as expected.

## Additional resources {#dev-get-start-ai-development-best-practices__additional-resources}

See the following additional resources for more information about prompting AI.

* [General guidelines for writing instructions for generative AI large language models (LLMs)](https://www.servicenow.com/docs/access?context=llm-instruction-guidelines&version=australia&pubname=australia-intelligent-experiences&ft:locale=en-US)
* [General guidelines for using app generation](https://servicenow-prod.fluidtopics.net/zo8fuhKg6AJr2aWT~39iMQ "To get the best results from app generation, provide clear and detailed descriptions of your application requirements during your conversation with ServiceNow Otto.")
* [Now Assist for Creator skill and prompting guides](https://www.servicenow.com/community/now-assist-for-creator-articles/now-assist-for-creator-skill-and-prompting-guides/ta-p/3125858)