Hi All,
I have created a new AI skill object for text generation in the Appian Dev environment. The skill was successfully created and tested with different temperature settings, and it works as expected. I have integrated it with my process model, published, and saved the model.
However, when attempting to deploy the AI skill to the Test environment, the deployment fails with an import error. During deployment, there are no warnings or dependency issues shown. The deployment log indicates the following message:
Problems (1):aiSkill 8292bde1-7e28-4b70-a94f-890b679ab860 "NH_Generative_AI_Skill": An error occurred while creating aiSkill [uuid=8292bde1-7e28-4b70-a94f-890b679ab860]: Name is insufficiently unique (APNX-1-4071-032) (APNX-1-4071-007)
I have attempted to resolve the issue by changing the name of the AI skill, but the problem persists. Could someone assist me in identifying the potential cause of this issue?
Additionally, out of curiosity, could you let me know which LLM model Appian uses for generative AI?
Thanks
Kundan
Discussion posts and replies are publicly visible
Did you check for any existing AI skills on TEST? If you follow your development & deployment process, there should not be any objects which also does not exist on DEV.
Hi Stefan,
I have checked in the test, no existing AI skills there.
Could it be that you do not have enough permission to see all objects on TEST?
Thanks, Stefan.Got the actual reason for this issue from the Appian support team as below:-I have looked into this further and I see that Generative AI features are not enabled on your test site which is why the object cannot be deployed.At its core, the error message seen is rather misleading. When AI Skills fail to deploy for some reason (for example, the feature is not available/enabled on the target site, or not supported in that site's region) the process logs an error saying "The Name is insufficiently unique" as the default response for this type of error. We're actively discussing with our internal product teams on ways that we can improve this behavior to make this more clearly explain the actual issue.