The Zurich release has arrived! Interested in new features and functionalities? Click here for more

Sharon_Barnes
ServiceNow Employee
ServiceNow Employee

Prompt Smarter, Not Harder: A Developer’s Guide to AI-Powered Test Generation in ServiceNow 

 

If you’ve ever lost a weekend to manual test setup before a ServiceNow upgrade, you’re not alone. The release of Now Assist Test Generation in the Yokohama release brings a powerful new way to automate and scale your testing, without writing a single line of code. 

In our latest Platform Academy session, Yinka walks us through how natural language prompting is revolutionizing ATF (Automated Test Framework) in ServiceNow. 

 

 

Why This Matters 

 

Manual test creation is one of the most time-consuming, error-prone parts of the dev lifecycle. Whether it’s maintaining up-to-date test coverage or onboarding new QAs, the challenges stack up. That’s where Now Assist comes in, reducing test creation time by up to 80%. 

 

Key Takeaways 

 

  • Prompt Engineering Meets ServiceNow: Take your knowledge from ChatGPT and apply it to ATF test generation using natural language. 
  • Best Practices for Prompting: 
  • Be clear and concise. 
  • Provide relevant context (e.g., user roles, expected behavior). 
  • Specify desired outcomes (e.g., error messages, successful form submission). 
  • Use ServiceNow-native keywords like “validate”, “navigate”, “visible”. 
  • Limitations to Know: Currently, Now Assist doesn’t support custom UI steps, related lists, service portal forms, or complex scripting—but improvements are on the roadmap. 
  • Licensing: You’ll need Creator Pro Plus, and this feature is not available on personal developer instances. 
  • Adoption Tips: 
  • Start with a small pilot team. 
  • Focus on specific projects. 
  • Train your citizen developers. 
  • Integrate feedback and grow from there. 

 

The Road Ahead 

While Now Assist test generation currently focuses on new test creation, there are plans to support editing existing tests and even generating tests directly from user stories in the backlog. Your feedback is crucial. Head to the Idea Portal to vote on features you want to see.