Specialist Paper 1

Leadership needs to populate a dynamic form field with a summary or description created by a large language model (LLM) to facilitate more productive conversations with customers. Leadership also wants to keep a human in the loop to be considered in their AI strategy. Which prompt template type should the AI Specialist recommend?

• Sales Email

• Field Generation

• Record Summary

Universal Containers is considering leveraging the Einstein Trust Layer in conjunction with Einstein Generative AI Audit Data. Which audit data is available using the Einstein Trust Layer?

• Response accuracy and offensiveness score

• Hallucination score and bias score

• Masked data and toxicity score

Universal Containers wants to make a sales proposal and directly use data from multiple unrelated objects (standard and custom) in a prompt template. What should the AI Specialist recommend?

• Create a Flex template to add resources with standard and custom objects as inputs.

• Create a prompt template passing in a special custom object that connects the records temporarily.

• Create a prompt template-triggered flow to access the data from standard and custom objects.

What is an AI Specialist able to do when the “Enrich event logs with conversation data" setting in Einstein Copilot is enabled?

• View the user click path that led to each copilot action.

• View session data including user Input and copilot responses for sessions over the past 7 days.

• Generate details reports on all Copilot conversations over any time period.

Universal Containers’ current AI data masking rules do not align with organizational privacy and security policies and requirements. What should an AI Specialist recommend to resolve the issue?

• Enable data masking for sandbox refreshes.

• Configure data masking in the Einstein Trust Layer setup.

• Add new data masking rules in LLM setup.

An administrator wants to check the response of the Flex prompt template they've built, but the preview button is greyed out. What is the reason for this?

• The records related to the prompt have not been selected.

• The prompt has not been saved and activated.

• A merge field has not been inserted in the prompt.

Universal Containers’ data science team is hosting a generative large language model (LLM) on Amazon Web Services (AWS). What should the team use to access externally-hosted models in the Salesforce Platform?

• Model Builder

• App Builder

• Copilot Builder

An AI Specialist built a Field Generation prompt template that worked for many records, but users are reporting random failures with token limit errors. What is the cause of the random nature of this error?

• The number of tokens generated by the dynamic nature of the prompt template will vary by record.

• The template type needs to be switched to Flex to accommodate the variable amount of tokens generated by the prompt grounding.

• The number of tokens that can be processed by the LLM varies with total user demand.

An administrator is responsible for ensuring the security and reliability of Universal Containers' (UC) CRM data. UC needs enhanced data protection and up-to-date AI capabilities. UC also needs to include relevant information from a Salesforce record to be merged with the prompt. Which feature in the Einstein Trust Layer best supports UC's need?

• Data masking

• Dynamic grounding with secure data retrieval

• Zero-data retention policy

A Salesforce Administrator is exploring the capabilities of Einstein Copilot to enhance user interaction within their organization. They are particularly interested in how Einstein Copilot processes user requests and the mechanism it employs to deliver responses. The administrator is evaluating whether Einstein Copilot directly interfaces with a large language model (LLM) to fetch and display responses to user inquiries. How does Einstein Copilot handle user requests in Salesforce?

• Einstein Copilot will trigger a flow that utilizes a prompt template to generate the message.

• Einstein Copilot will perform an HTTP callout to an LLM provider.

• Einstein Copilot analyzes the user's request and LLM technology is used to generate and display the appropriate response.