I have created New AI agent from AI agent Studio with below Use case but results are not consistent and it is not working as expected.
Use Case: Table 'x_**xyz' have fields Topic, Description and Primary Attachment (file attachment type field)
Whenever new Record created with Primary Attachment the AI agent should trigger to Run Semantic Comparison with All other records from same table considering Topic, Description and Attachment Data, This is to get Similarity score to decide if similar record submitted already so that Approval can review accordingly while approving record. I build multiple Tools within AI agent to provide Input like Flow Action and Sub flow for below activity:
1. Flow Action to get Current record Number, Topic, Description and Attachment data
2. Flow Action to get Comparison candidates as Array of sys_id's
3. Flow Action to get Secondary comparison data Number, Topic, Description and attachment data for each Sys id
4. Subflow to Update Comparison Summary and score in current record.
When I tried to test AI agent the tools are executing correctly with correct input and output
but in prompt I mentioned for each Comparison candidate comparison should execute (Here it working for few randomly but skipped for others)
and score also picked randomly from any of comparison.
Do we have any better solution where no candidate can skip and AI agent can return consistence response?
