Salesforce Agentforce Specialist Certification Study

Vocabulary practice flashcards covering key concepts from the Salesforce Agentforce Specialist Certification transcript, including AI fundamentals, Einstein Trust Layer features, and Prompt Builder configurations.

Plan Canvas

Where an AI Specialist goes to access user interactions with the Agent, including errors, incorrectly triggered actions, and incomplete plans.

CRM Data Grounding

A technique that ensures AI-generated responses are based on real-time, accurate data from Salesforce records using merge fields.

Numeric Prediction

An AI category used to forecast a numerical value based on historical data, such as determining the optimal price of an airline ticket.

Structured Data

Data organized and formatted in a searchable way, such as a database containing business names, zip codes, and market values.

Deep Learning

A type of AI training involving neural networks with multiple layers that allow the model to learn complex patterns.

Natural Language Processing (NLP)

The field of AI focused on enabling machines to understand, interpret, and respond to everyday human language.

Prediction

In the context of generative AI, the process where the model forecasts a sequence of words likely to have meaning and relevance to the reader.

Dynamic Actions

A Salesforce feature that allows admins to set filters based on user fields so specific actions are visible only to certain users.

Intelligence View

A view available for leads, contacts, and accounts that provides key metrics and actions in one place.

Pearson's Correlation

A technique that shows whether and how strongly pairs of quantitative variables are related, without proving causation.

Prompt Defense

System policies within the Einstein Trust Layer designed to limit hallucinations and decrease the likelihood of unintended or harmful outputs.

Data Masking

A feature of the Einstein Trust Layer that detects and replaces sensitive information with placeholder text before it is sent to the Large Language Model.

Zero-Data Retention Policy

A policy ensuring that no patient or customer data is used for LLM model training or product improvements.

Einstein Lead Scoring

An AI feature used to score and prioritize leads based on how well they fit a company's successful conversion patterns.

Einstein Service Replies for Email

A Service Cloud feature that generates personalized email responses to customers based on recommended Knowledge articles.

Einstein Work Summaries

A feature that predicts and fills out a summary, issue, and resolution following customer conversations.

Flow Merge Fields

A grounding technique used in Prompt Builder to incorporate complex logic and real-time data into a prompt template.

Flex Prompt Template

A versatile template in Prompt Builder used for various business purposes that allows the definition of custom resources and inputs.

Field Generation Prompt Template

A template type used to populate content for record fields in Lightning Experience, requiring the page to be upgraded to Dynamic Forms.

Toxicity Scoring

A tool that evaluates content for toxicity and logs scores in Data Cloud as part of an audit trail.

Call Summaries

A Sales Cloud feature for creating and sharing editable summaries of voice and video calls.

Prompts

Instructions given to a language model that guide it in generating specific content according to desired requirements.

Goal (in a prompt)

The specific outcome or desired objective a user hopes to achieve with the model's output.

Agentforce Testing Center

A tool where running tests does not consume Einstein Requests and does not risk modifying CRM data in production.

Einstein Copilot

An AI-powered virtual assistant integrated into Salesforce that offers contextual guidance and automation to optimize productivity.

Event Logs (in Copilot Builder)

Logs used by administrators to monitor and troubleshoot a copilot's conversation activity with users.

Copilot Planner Service

The component that orchestrates how a copilot launches actions during a conversation to accomplish a user task.

Copilot Analytics Dashboard

A dashboard showing two types of information: User Data (engagement and preferences) and Action Data (execution and success rates).