Agentforce — Salesforce Platform Administrator practice questions

Which sales feature gives reps AI-assisted guidance and forecasting?

1. The Service Console, which arranges cases and knowledge for support agents
2. A developer-built scoring model deployed as a custom page component
3. Marketing Cloud, which runs email journeys aimed at prospects and customers
4. Einstein for Sales: opportunity and lead scoring, forecasting and insights

Answer: D — Einstein for Sales: opportunity and lead scoring, forecasting and insights

D) Right. Einstein for Sales supplies opportunity and lead scoring, forecasting insights and activity capture inside the standard sales objects, configured by the admin rather than coded. A) The Service Console is a service workspace. C) Marketing Cloud is a separate product for campaigns. B) Custom development is not the administrator's path to these features.

What does the Einstein Trust Layer's zero data retention policy with supported model providers mean?

1. Salesforce deletes the org's CRM records once each AI request has finished being processed by the model
2. The provider stores prompts for a fixed retention window of about a month and then purges them in bulk
3. Audit trail capture of AI activity is switched off by default and must be enabled by the administrator
4. The provider retains no prompts or responses and cannot train its models on the customer's data

Answer: D — The provider retains no prompts or responses and cannot train its models on the customer's data

D) Correct. Under the agreements behind the trust layer, the external provider keeps neither the prompt nor the response and may not use customer data for training. A) CRM records are untouched by the policy. B) The point is no retention, not a shorter window. C) Audit trail capture remains a separate, available trust layer feature.

An agent returns account information but consistently omits one field that users expect. Permissions on the object are confirmed. Which cause is MOST likely?

1. The field is hidden from the agent user by field-level security
2. The record is owned by a user outside the agent user's role branch
3. The org-wide default for the object is set to Private for internal users
4. The field is a formula, and formulas are never available to any agent

Answer: A — The field is hidden from the agent user by field-level security

A) Correct — object access without field access produces exactly this symptom: records come back, one field does not. B) would withhold whole records rather than a single field, and object access was already confirmed. C) likewise affects record visibility, not individual fields. D) invents a restriction; formula fields are readable when field-level security allows.

What is Prompt Builder used for?

1. Writing and testing queries against the org's objects with autocomplete and a grid of returned rows
2. Creating reusable prompt templates that merge CRM record data into the instructions sent to the model
3. Building screen flows by dragging components onto screens and wiring the resources between them
4. Designing a record page layout, arranging the fields, related lists, and components on each of the app pages

Answer: B — Creating reusable prompt templates that merge CRM record data into the instructions sent to the model

B) Correct. Prompt Builder creates grounded, reusable prompt templates whose merge fields, related lists, and flows pull org data into the instruction sent to the model. A) Query tooling lives elsewhere. C) Screen flows are built in Flow Builder. D) Page composition is done in the Lightning App Builder.

When an Agentforce agent runs a flow action, whose data access applies in a properly configured deployment?

1. The permissions of the user context configured for the agent, so sharing and object access still apply
2. Full system administrator access at all times, regardless of how the agent's user context is configured
3. No access at all, because flow actions invoked by an agent are prevented from reading or writing data
4. The permissions of whichever user in the org logged in most recently before the agent ran the action

Answer: A — The permissions of the user context configured for the agent, so sharing and object access still apply

A) Correct. Agent actions execute in a defined user context, so profile, permission set, and sharing settings still govern what the agent can read and change. This is why troubleshooting agent permissions starts with that user. B) Unrestricted access would be a security failure. C) Actions exist to touch data. D) Access never depends on recent logins.

Why can an Agentforce agent fail to see or update a record, and how is it fixed?

An agent runs as a designated agent user, so it is bound by that user's profile, permission sets, object and field-level security, record sharing and org-wide defaults — not by the permissions of the person chatting with it. Fix it by granting the agent user access to the object, fields and records (permission set plus sharing), and by confirming the topic's action is actually assigned to the agent.

Einstein features?

AI/ML built into Salesforce. Lead scoring, opportunity insights, next best action, etc.

What is Agentforce and what does an admin configure for it?

Agentforce is Salesforce's platform for autonomous AI agents that act on CRM data. An admin defines the agent, its topics (the jobs it covers) and actions (what it can do — flows, prompt templates, Apex built by others, standard actions), sets its instructions and scope, assigns the agent user and permissions, and tests in Agent Builder's preview before activating.