Generative AI text-based fallback
- 최종 업데이트2026/01/14
If the object on which you want to execute an 작업 is not detected within the specified time-out, you can leverage generative AI based fallback to automate the specific dynamic webpages, decreasing the possibility of the 레코더 작업 failure in the 봇 due to UI element changes or dynamic properties.
Once the 생성형 레코더 verifies that a fallback DOMXPath can uniquely identify the target UI element, it conducts an additional validation step. This involves comparing other properties of the UI element. For instance, if the automated 작업 involves data entry, the 생성형 레코더 ensures that the automation proceeds only if the fallback DOMXPath corresponds to a data entry input, such as a text box, and not to buttons or labels.
- From the Bot 편집기, navigate to .
- In the , enable the Use Generative AI-based
fallback option.
- These are 봇 level settings. Hence, when you enable generative AI-based fallback, all the capture 작업 occurring in that specific 봇 will use these same settings with a time out of 30 seconds.
- Starting from 레코더 패키지 (version 3.1.5 and later), the 생성형 레코더 configuration settings will be visible regardless of whether a 레코더 작업 is added in the Bot 편집기. To save the configuration settings, ensure that you use at least one 작업 in your automation.
Automatically update DOMXpath values in the debug mode
To update these values at runtime, run your automations in Debug mode. You can choose to update your DOMXPath with the recommended value or to skip it. When you select the Stop automation option, the 봇 execution will be stopped.
When you update your DOMXPath with the recommended value, ensure that you deselect other invalid or dynamic properties in your search criteria before running your automations again. If these dynamic properties remain selected and are combined with the new DOMXPath, the Recorder fails and fallback is triggered again.
You can also move the fallback notification dialogue window at the desired location on the screen according to your needs and preferences thus improving the visibility of your business application and the fallback.
Caching of successful Generative AI responses
The Generative AI calls made during successful Generative AI-based fallbacks are now cached in the system memory and the region-based service which are then reused for subsequent executions of the same automation. This reduces the time taken for subsequent executions of the same automation significantly thus improving performance.