Trusted AI guidelines and the Einstein Trust Layer
Apply Salesforce's guidelines for trusted AI (Accuracy, Safety, Honesty, Empowerment, Sustainability) to sales scenarios such as unreviewed AI emails and score-based lead disqualification, recognize bias and red-teaming risk categories, and map security concerns to Einstein Trust Layer features, including the rule that data masking is disabled for agents.
Identify ethical challenges of AI and apply Salesforce's Trusted AI Principles to sales scenarios.
~7 min read · 11 practice questions ·
What the exam expects
- Name Salesforce's five guidelines for trusted AI and pick the one a sales scenario violates or needs.
- Recognize ethical risks in sales AI (inaccurate drafts, undisclosed AI content, bias inherited from historical data, automation bias) and recommend human-in-the-loop controls.
- List the risk categories Salesforce red-teams for and how it tests.
- Map a security or compliance concern to the Einstein Trust Layer feature that answers it, and know that data masking is disabled for agents.
Key terms: trusted AI guidelines (Accuracy, Safety, Honesty, Empowerment, Sustainability); human in the loop (a person reviews or decides before AI output takes effect); red-teaming ("a structured process for probing AI systems and products for the identification of harmful capabilities, outputs, or infrastructural threats"); Einstein Trust Layer (Salesforce's set of agreements, security technology and data and privacy controls for generative AI); zero data retention (third-party LLMs keep nothing and train on nothing).
Key facts
- Detail
- Accuracy, Safety, Honesty, Empowerment, Sustainability
- Detail
- Respect data sources, get consent, disclose AI-generated content
- Detail
- Humans make high-risk decisions; routine tasks automated
- Detail
- Toxicity, bias and security; manual and automated
- Detail
- Grounding uses the executing user's permissions and field-level security
- Detail
- Not retained, not used for training by third-party LLMs
- Detail
- Data masking for LLMs is disabled for agents
- Detail
- Prompts, responses, toxicity scores and trust signals in Data 360
- Detail
- Drafts must be reviewed and customized before sending
Sources for this section (5)
- Promote Responsible and Ethical Agents (Trailhead)trailhead.salesforce.com (opens in a new tab)
- Discover How Salesforce Builds Trusted Agentic AItrailhead.salesforce.com (opens in a new tab)
- Einstein Trust Layerhelp.salesforce.com (opens in a new tab)
- Set up Einstein Trust Layerhelp.salesforce.com (opens in a new tab)
- Considerations for Einstein Generative AI Sales Emailshelp.salesforce.com (opens in a new tab)
- Why is auto-disqualifying low-scoring leads an Empowerment concern?
- When does Trust Layer data masking not protect a prompt?
- Which three risk categories does Salesforce red-team AI for?
- Why can't a junior rep's email draft expose an opportunity field she can't see?
The questions are free to read. Answers with their sources and self-grading are included with the pack.
Keep reading with the Agentforce Sales Consultant pack
- Core concepts
- Exam traps
- Worked scenario
The pack includes:
- All 20 lessons of the Agentforce Sales Consultant guide, with reading progress
- Interactive decision walks and step-by-step sequence steppers
- Inline practice questions and one-click topic practice (11 questions for this lesson)
- An official source cited for every claim
- An AI tutor that answers from the guide
- Unlimited timed practice exams