Sample Questions Flashcards
You are a data scientist who needs to view and manage models in Einstein Studio. You also need to create prompt templates in Prompt Builder.
Which permission sets should you assign to the data scientist?
1. Prompt Template User and Data Cloud Admin
2. Prompt Template Manager and Prompt Template User
3. Data Cloud Admin and Prompt Template Manager
- Data Cloud Admin and Prompt Template Manager
You are considering using a Field Generation prompt template type.
What should you check before creating the Field Generation prompt to ensure it is possible for the field to be enabled for generative AI?
- That the field chosen must be a rich text field with 255 characters or more
2. The Lightning page layout where the field will reside has been upgraded to Dynamic Forms
3. That the org is set to API version 59 or higher
2. The Lightning page layout where the field will reside has been upgraded to Dynamic Forms
you configured Data Masking within the Einstein Trust Layer.
How should you begin validating that the correct fields are being masked?
- Request the Einstein Generative AI Audit Data from the Security section of the Setup menu.
- Enable the collection and storage of Einstein Generative Al Audit Data on the Einstein Feedback setup page.
- Use a Flow-based resource in Prompt Builder to debug the fields’ merge values using Flow Debugger.
- Enable the collection and storage of Einstein Generative Al Audit Data on the Einstein Feedback setup page.
You want to create a new Sales Email prompt template in Prompt Builder using the “Save As” function. However, UC notices that the new template produces different results compared to the standard Sales Email prompt due to missing hyperparameters.
What should UC do to ensure the new prompt template produces results comparable to the standard Sales Email prompts?
- Revert to using the standard template without modifications.
2. Manually add the hyperparameters to the new template
3. Use Model Playground to create a model configuration with the specified parameters.
3. Use Model Playground to create a model configuration with the specified parameters.
Your current AI data masking rules do not align with organizational privacy and security policies and requirements.
What should you recommend to resolve the issue?
1. Configure data masking in the Einstein Trust Layer setup
2. Add new data masking rules in LLM setup.
3. Enable data masking for sandbox refreshes.
- Configure data masking in the Einstein Trust Layer setup
You want to assess Salesforce’s generative AI features but have concerns over its company data being exposed to third-party large language models (LLMs). Specifically, UC wants the following capabilities to be part of Einstein’s generative Al service.
- No data is used for LLM training or product improvements by third-party LLMs.
- No data is retained outside of UC’s Salesforce org.
- The data sent cannot be accessed by the LLM provider.
Which property of the Einstein Trust Layer should you highlight to UC that addresses these requirements?
1. Data Masking
2. Zero-Data Retention Policy
3. Prompt Defense
2. Zero-Data Retention Policy
You turned on Einstein Generative AI in Setup. Now, you would like to create custom prompt templates in Prompt Builder. However, they cannot access Prompt Builder in the Setup menu.
What is causing the problem?
1. The Prompt Template Manager permission set was not assigned correctly.
2. The large language model (LLM) was not configured correctly in Data Cloud.
3. The Prompt Template User permission set was not assigned correctly.
- The Prompt Template Manager permission set was not assigned correctly.
You 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?
1. The template type needs to be switched to Flex to accommodate the variable amount of tokens generated by the prompt grounding.
2. The number of tokens that can be processed by the LLM varies with total user demand.
3. The number of tokens generated by the dynamic nature of the prompt template will vary by record.
- The number of tokens generated by the dynamic nature of the prompt template will vary by record.
You are working on a prompt template to generate personalized emails for product demonstration requests from customers. It is important for the AI-generated email to adhere strictly to the guideline a, using only associated opportunity information, and to encourage the recipient to take the desired action.
How should you include these instructions on a new line in the prompt template?
1.Use curly brackets () to encapsulate instructions
2. Make sure merged fields are defined.
3. Surround them with triple quotes (“””).
- Surround them with triple quotes (“””).
Which feature in the Einstein Trust Layer helps to minimize the risks of jailbreaking and prompt injection attacks?
- Data Masking
2. Secure Data Retrieval and Grounding
3. Prompt Defense
3. Prompt Defense
Your service team wants to customize the standard case summary response from Einstein Copilot.
What should you do to achieve this?
1.Customize the standard Record Summary template for the Case object.
2. Summarize the Case with a standard copilot action.
3. Create a custom Record Summary prompt template for the Case object.
1 or 3
You plan to enhance the customer support team’s productivity using AI.
Which specific use case necessitates the use of Prompt Builder?
- Creating a draft of a support bulletin post for new product patches
- Estimating support ticket volume based on historical data and seasonal trends
- Creating an AI-generated customer support agent performance score
- Creating a draft of a support bulletin post for new product patches
A service agent is looking at a custom object that stores travel information. They recently received a weather alert and now need to cancel flights for the customers that are related with this itinerary. The service agent needs to review the Knowledge articles about canceling and rebooking the customer flights.
Which Einstein Copilot capability helps the agent accomplish this?
1. Invoke a flow which makes a call to external data to create a Knowledge article
2. Execute tasks based on available actions, answering questions using information from accessible Knowledge articles.
3. Generate a Knowledge article based off the prompts that the agent enters to create steps to cancel flights.
2. Execute tasks based on available actions, answering questions using information from accessible Knowledge articles.
You have a legacy system that needs to integrate with Salesforce. UC wishes to create a digest of account action plans using the generative API feature.
Which API service should UC use to meet this requirement?
- SOAP API
2. REST API
3. Metadata API
2. REST API
You want to use the related lists from an account in a custom prompt template.
What should you consider when configuring the prompt template?
- The text encoding (for example, UTF-8, ASCII) option
2. The choice between XML and JSON rendering formats for the list - The maximum number of related list merge fields
- The maximum number of related list merge fields
How does the Einstein Trust Layer ensure that sensitive data is protected while generating useful and meaningful responses?
1. Masked data will be de-masked during response journey.
2. Masked data will be de-masked during request journey.
3. Responses that do not meet the relevance threshold will be automatically rejected
- Masked data will be de-masked during response journey.
You have implemented Generative Al within Salesforce to enable summarization of a custom object called Guest. Users have reported mismatches in the generated information.
In refining its prompt design strategy, which key practices should UC prioritize?
1. Create concise, clear, and consistent prompt templates with effective grounding, contextual role-playing, clear instructions, and iterative feedback.
2. Submit a prompt review case to Salesforce and conduct thorough testing in the playground to refine outputs until they meet user expectations.
3. Enable prompt test mode, allocate different prompt variations to a subset of users for evaluation, and standardize the most effective model based on performance feedback.
1. Create concise, clear, and consistent prompt templates with effective grounding, contextual role-playing, clear instructions, and iterative feedback.
you of Northern Trail Outfitters reviewed the organization’s data masking settings within the Configure Data Masking menu within Setup. Upon assessing all of the fields, a few additional fields were deemed sensitive and have been masked within Einstein’s Trust Layer.
Which steps should you take upon modifying the masked fields?
- Turn on Einstein Feedback so that end users can report if there are any negative side effects on Al features.
2. Test and confirm that the responses generated from prompts that utilize the data and masked data do not adversely affect the quality of the generated response
3. Turn off the Einstein Trust Layer and turn it on again.
2. Test and confirm that the responses generated from prompts that utilize the data and masked data do not adversely affect the quality of the generated response
You are considering leveraging the Einstein Trust Layer in conjunction with Einstein Generative AI Audit Data.
Which audit data is available using the Einstein Trust Layer?
- Hallucination score and bias score
- Masked data and toxicity score
- Response accuracy and offensiveness score
- Masked data and toxicity score
What is the main purpose of Prompt Builder?
- A tool for developers to use in Visual Studio Code that creates prompts for Apex programming, assisting developers in writing code more efficiently.
- A tool that enables companies to create reusable prompts for large language models (LLMs), bringing generative Al responses to their flow of work
- A tool within Salesforce offering real-time Al-powered suggestions and guidance to users, improving productivity and decision-making.
2. A tool that enables companies to create reusable prompts for large language models (LLMs), bringing generative Al responses to their flow of work
You want to be able to detect with a high level of confidence if content generated by a large language model (LLM) contains toxic language.
Which action should you take in the Trust Layer to confirm toxicity is being appropriately managed?
- Create a Trust Layer audit report within Data Cloud that uses a toxicity detector type filter to display toxic responses and their respective scores.
2. Create a flow that sends an email to a specified address each time the toxicity score from the response exceeds a predefined threshold.
3. Access the Toxicity Detection log in Setup and export all entries where istoxicityDetected is true.
- Create a Trust Layer audit report within Data Cloud that uses a toxicity detector type filter to display toxic responses and their respective scores.
You want to allow its service agents to query the current fulfillment status of an order with natural language. There is an existing autolaunched flow to query the information from Oracle ERP, which is the system of record for the order fulfillment process.
How should you apply the power of conversational AI to this use case?
1. Create a custom copilot action which calls a flow.
2. Configure the Integration Flow Standard Action in Einstein Copilot.
3. Create a Flex prompt template in Prompt Builder.
1. Create a custom copilot action which calls a flow.
You need to create a Sales Email with a custom prompt template. They need to ground on the following data.
Opportunity Products
Events near the customer
Tone and voice examples
How should you obtain related items?
- Call a prompt initiated flow to fetch and ground the required data.
2. Utilize a standard email template and manually insert the required data fields.
3. Create a flex template that takes the records in question as inputs.
- Call a prompt initiated flow to fetch and ground the required data.
You implement Custom Copilot Actions to enhance its customer service operations. The development team needs to understand the core components of a Custom Copilot Action to ensure proper configuration and functionality.
What should the development team review in the Custom Copilot Action configuration to identify one of the core components of a Custom Copilot Action?
1. Instructions
2. Output Types
3. Action Triggers
- Instructions
You are tasked with configuring a generative model to create personalized sales emails using customer data stored in Salesforce. You have already fine-tuned a large language model (LLM) on the OpenAI platform. Security and data privacy are critical concerns for the client.
How should you integrate the custom LLM into Salesforce?
1. Add the fine-tuned LLM in Einstein Studio Model Builder.
2. Create an application of the custom LLM and embed it in Sales Cloud via Frame.
3. Enable model endpoint on OpenAl and make callouts to the model to generate emails.
1. Add the fine-tuned LLM in Einstein Studio Model Builder.
You have created a copilot custom action using flow as the reference action type. However, it is not delivering the expected results to the conversation preview, and therefore needs troubleshooting.
What should you do to identify the root cause of the problem?
1. In Copilot Builder within the Dynamic Panel, confirm selected action and observe the values in Input and Output sections.
2. In Copilot Builder, verify the utterance entered by the user and review session event logs for debug information.
3. In Copilot Builder within the Dynamic Panel, turn on dynamic debugging to show the inputs and outputs.
3. In Copilot Builder within the Dynamic Panel, turn on dynamic debugging to show the inputs and outputs.
A sales rep at You is extremely busy and sometimes will have very long sales calls on voice and video calls and might miss key details. They are just starting to adopt new generative AI features.
Which Einstein Generative Al feature should you recommend to help the rep get the details they might have missed during a conversation?
- Call Explorer
2. Call Summary
3. Sales Summary
2. Call Summary
You are experimenting with using public Generative Al models and are familiar with the language required to get the information it needs. However, it can be time consuming for both UC’s sales and service reps to type in the prompt to get the information they need, and ensure prompt consistency.
Which Salesforce feature should you recommend to address these concerns?
- Einstein Recommendation Builder
2. Einstein Prompt Builder and Prompt Templates
3. Einstein Copilot Action: Query Records
2. Einstein Prompt Builder and Prompt Templates
What is best practice when refining Einstein Copilot custom action instructions?
- Use consistent introductory phrases and verbs across multiple action instructions.
2. Provide examples of user messages that are expected to trigger the action.
3. Specify the persona who will request the action.
2. Provide examples of user messages that are expected to trigger the action.
When configuring a prompt template, you preview the results of the prompt template they’ve written. They see two distinct text outputs: Resolution and Response.
Which information does the Resolution text provide?
- It shows the full text that is sent to the Trust Layer.
2. It shows which sensitive data is masked before it is sent to the LLM
3. It shows the response from the LLM based on the sample record.
2. It shows which sensitive data is masked before it is sent to the LLM
You recently launched a pilot program to integrate conversational Al into its CRM business operations with Einstein Copilot.
How should you monitor Copilot’s usability and the assignment of actions?
- Run Einstein Copilot Analytics.
2. Run a report on the Platform Debug Logs.
3. Query the Copilot log data using the metadata API.
- Run Einstein Copilot Analytics.
You (UC) recently rolled out Einstein Generative Al capabilities and has created a custom prompt to summarize case records. Users have reported that the case summaries generated are not returning the appropriate information.
What is a possible explanation for the poor prompt performance?
- The Einstein Trust Layer is incorrectly configured.
2. The data being used for grounding is incorrect or incomplete.
3. The prompt template version is incompatible with the chosen LLM.
2. The data being used for grounding is incorrect or incomplete.
What is the primary function of the planner service in the Einstein Copilot system?
1. Generating record queries based on conversation history
2. Identifying copilot actions to respond to user utterances
3. Offering real-time language translation during conversations
2. Identifying copilot actions to respond to user utterances
You is very concerned about security compliance and wants to understand:
Which prompt text is sent to the large language model (LLM)
How it is masked
The masked response
What should you recommend?
- Ingest the Einstein Shield Event logs into CRM Analytics.
2. Enable audit trail in the Einstein Trust Layer.
3. Review the debug logs of the running user.
2. Enable audit trail in the Einstein Trust Layer.
You (UC) noticed an increase in customer contract cancellations in the last few months. UC is seeking ways to address this issue by implementing a proactive outreach program to customers before they cancel their contracts and is asking the Salesforce team to provide suggestions.
Which use case functionality of Model Builder aligns with UC’s request?
1. Customer churn prediction
2. Contract Renewal Date prediction
3. Product recommendation prediction
1. Customer churn prediction
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.
- A merge field has not been inserted in the prompt.
- The prompt has not been saved and activated.
- The records related to the prompt have not been selected.
What is the correct process to leverage Prompt Builder in a Salesforce org?
- Enable the target object for generative prompting, develop the prompt within the prompt workspace, select records to fine-tune and ground the response, enable the Trust Layer, and associate the prompt to an action.
- Select the appropriate prompt template type to use, select one of Salesforce’s standard prompts, determine the object to associate the prompt, select a record to validate against, and associate the prompt to an action.
- Select the appropriate prompt template type to use, develop the prompt within the prompt workspace, select resources to dynamically insert CRM-derived grounding data, pick the model to use, and test and validate the generated responses.
3. Select the appropriate prompt template type to use, develop the prompt within the prompt workspace, select resources to dynamically insert CRM-derived grounding data, pick the model to use, and test and validate the generated responses.