Configure Business Component User Properties for Multimodal AI

To configure Business Component User Properties: Use Case 1

The above configurations for RunTime Event, BC User properties, Applet/View/Screen/Link/BO configurations are given for the OOTB use case. Please configure the same accordingly for your use cases.

1. Required Seed Data changes:

   1. Click on Sitemap.
   2. Navigate to Administration- Application > Views.
   3. Navigate to Generative AI Configuration List View and create a new record as per the following table:

      Prompt Name	Prompt Value
      ClosingSummaryPrompt	Given below is a transcript of a customer support interaction. The customer support interaction can contain an attached image or document as well. Please holistically summarize the entire text and the image data. The summary should highlight the main issue, the exact device or account details, and the actions undertaken to resolve it. Do mandatorily summarize the image or document if present. Ensure that the summary maintains the contextual history accurately and does not infer or create any conclusions about the customer's state of mind or satisfaction unless explicitly stated by the customer. Strictly avoid adding any information that is not clearly present in the interaction. Please look for the image or document attachment. The textual data is given below:
      TransferSummaryPrompt	Given below is a transcript of a customer support interaction. The customer support interaction can contain an attached image or document as well. Please holistically summarize the entire text and the image data. The summary should highlight the main issue, the remaining or pending concern, the exact device or account details, and the actions undertaken to resolve it. Do mandatorily summarize the image or document if present. Ensure that the summary maintains the contextual history accurately and does not infer or create any conclusions about the customer's state of mind or satisfaction unless explicitly stated by the customer. Strictly avoid adding any information that is not clearly present in the interaction. Please look for the image or document attachment.

2. Required Repository Changes:

   1. Create Business Component User Properties.

   2. Add following Business Component User Properties while using the LLAMA (meta.llama-XXXX) LLM:

      BC User Property Name	BC User Property Value
      SummarizeChatText_Param_1	compartmentId: <CHANGE_ME> [Pick the value from OCI tenancy or OCI configuration file]
      SummarizeChatText_Param_2	servingMode:ON_DEMAND [Possible values are ON_DEMAND or DEDICATED]
      SummarizeChatText_Param_3	lovInfo:BC_FIELD=Type;LOV_TYPE=TODO_TYPE;LOV_LIC=CLOSING SUMMARY,TRANSFER SUMMARY [This parameter is used to pick the appropriate Localization value]
      SummarizeChatText_Param_4	input:FIELD=Description [This is the BC field whose input value will be send to the LLM]
      SummarizeChatText_Param_5	inputFromOtherBC:BUSCOMP=Service Request,FIELD=Abstract;BUSCOMP=Order Entry - Orders,FIELD=Order Priority,Account [This is the list of other BC fields whose input values will be send to the LLM along with that of the primary BC. Please note that all the other BC's should be part of the same BO ]
      SummarizeChatText_Param_6	output:FIELD=Description;MAX_FIELD_LENGTH=250 [This is the BC field to which the LLM response will be populated. Please modify the MAX_FIELD_LENGTH based on the respective BC Field length]
      SummarizeChatText_Param_7	prompt:CLOSING SUMMARY=ClosingSummaryPrompt;TRANSFER SUMMARY=TransferSummaryPrompt [This is the input prompt to the LLM which is/are defined in the Administration - Generative AI screen]
      SummarizeChatText_Param_8	minimumInputCharLength:250 [This is the minimum input length to the LLM]
      SummarizeChatText_Param_9	runtimeType:LLAMA [This is the minimum input length to the LLM]
      SummarizeChatText_Param_10	imageUrl:BUSCOMP=Service Request Attachment,FIELD=ActivityFileName,SORT_SPEC=ActivityFileDate(DESC) [This is the input Field to specify the Image Attachment to the LLM]

      Note: SummarizeChatText_Param (1-8) are available out of the box. Whereas, SummarizeChatText_Param (9-10) requires configuration.
3. Modify ociconfig.properties file, make the following entry in the [GENAI] section:

   multimodal_image_type = png,jpg,jpeg
   multimodal_max_image_size = 5242880
   multimodal_image_for_summarize = true

To configure Business Component User Properties: Use Case 2

1. Required Seed Data changes:

   1. Click on Sitemap.
   2. Navigate to Administration- Application > Views.
   3. Navigate to Generative AI Configuration List View and create a new record as per the following table:

      Prompt Name

      Prompt Value

      ExtractImagePrompt

      Analyze the provided image and extract the address details. Extract the mentioned Address, City, State, Zip Code and the Country from the image. Do not extract any information which is not present in the image. The response should contain only the address, city, state, zip code and the country details. Display the attribute name, followed by a colon and its value and then leave a line space after each attribute. Do not include any other text in the response.

2. Required Repository Changes:

   1. Create Business Component User Properties for Service Request Attachment.

      BC User Property Name	BC User Property Value
      ExtractImageChat_Param_1	compartmentId: <CHANGE_ME> [Pick the value from OCI tenancy or OCI configuration file]
      ExtractImageChat_Param_2	servingMode:ON_DEMAND [Possible values are ON_DEMAND or DEDICATED]
      ExtractImageChat_Param_3	prompt:ExtractImagePrompt [This is the input prompt to the LLM which is/are defined in the Administration - Generative AI screen]
      ExtractImageChat_Param_4	imageUrl:BUSCOMP=Service Request Attachment,FIELD=ActivityFileName [This is the BC field whose input value will be send to the LLM]
      ExtractImageChat_Param_5	runtimeType:LLAMA [Type or provider of the LLM like LLAMA]
      ExtractImageChat_Param_6	outputToOtherBC:BUSCOMP=GenAI Image Extract Response BC,FIELD=Address Maint - New Addr Line 1:Address,Address Maint - New City:City,Address Maint - New Country:Country,Address Maint - New State:State,Address Maint - New Zip Code:Zip Code; [This user property is used when we need to populate the output to other BC, apart from the current BC.]. FIELD=<BC Field>:<Param from LLM Response>
      Named Method 1	"Extract", "INVOKESVC", "Service Request Attachment", "SiebelOCIAIService", "InvokeOCIAI", "'Context'", "'GenAI:ExtractImageChat'" [This is the BS Method invoked on click of the Button to perform functionality like Extracting content from an image]

3. Modify ociconfig.properties file, make the following entry in the [GENAI] section:

   multimodal_image_type = png,jpg,jpeg
   multimodal_max_image_size = 5242880
   multimodal_image_for_summarize = true