Explanation:

Summary:
Einstein Agent components, including the agent configuration, topics, and actions, are fully supported by Salesforce's standard deployment tools. The most straightforward and recommended method for moving metadata from a sandbox to production is by using Change Sets, which provide a user-friendly interface for selecting and deploying components without needing code.

Correct Option:

C) 📤 Create an outbound change set with all the necessary agent components, then upload to production.
🔹 Change Sets are the standard, no-code deployment tool for moving customizations between Salesforce environments.
🔹 Einstein Agent components are packaged as metadata and can be added to a Change Set just like any other customizable feature (e.g., custom objects, fields).
🔹 This method is administered directly within the Salesforce Setup menu, making it the most accessible option for administrators.

Incorrect Option:

A) 🔄 Manually recreate the agent... because change sets cannot be used.
This is incorrect. Agent components are deployable metadata. Manually recreating them is time-consuming, error-prone, and unnecessary.

B) 💾 Export agent components as JSON files and manually import them... using the Metadata API.
While technically possible, this is a complex, developer-oriented approach. Using the Metadata API directly is not the standard administrative path when Change Sets are available and fully supported.

Reference
Salesforce Help: Components Available in Change Sets

Explanation:

🧭 Summary:
When an Agent “hallucinates” weblinks or other irrelevant content, it means the model generated output not grounded in retrieved data. Since this Agent uses a prompt template with a knowledge retriever, the issue likely lies in how retrieved content and instructions are presented within the prompt. Reviewing the final resolved prompt and retrieved chunks reveals whether the model had accurate context to generate the response.

✅ Correct Option:

🟩 B. Examine the prompt instructions and contents of the chunks shown in the resolved prompt output.
This approach helps trace whether hallucinations come from unclear instructions or irrelevant retrieved content. The resolved prompt output shows what data the Agent actually received, enabling diagnosis of poor retrieval or confusing guidance that may have led to hallucinated weblinks.

❌ Incorrect Options:

🟥 A. Examine the topic name and classification description for hallucination guardrails.
The topic name and classification guardrails don’t directly affect hallucinations. They help route the query to the right process but won’t expose issues within the retrieval or generation logic that cause fabricated content.

🟥 C. Examine the topic instructions and ensure the word "ALWAYS" is used in the hallucination guardrails.
The presence or absence of the word “ALWAYS” won’t stop hallucinations. It’s about content grounding, not rule enforcement wording. Overemphasis on phrasing misses the deeper cause—poorly aligned retrieval or prompt context.

📘 Reference:
Salesforce Help: Troubleshoot Agent Hallucinations
Salesforce Help: Use Prompt Templates with Knowledge Retrieval

Explanation:

🧭 Summary:
The issue stems from a misunderstanding of what Salesforce Agent supports. The Agent in Salesforce can perform various record-related actions—like creating or updating records—based on configured permissions and flows. However, not every type of record manipulation is available. In this case, deletion is specifically not supported as an Agent action, which is why activities from the past 7 days aren’t being removed.

✅ Correct Option:

🟩 C. Agent does not support the Delete Record action.
Salesforce Agent currently doesn’t allow record deletion actions in its supported functionality. Even if the user has the right permissions, the Agent platform itself doesn’t execute record deletions, ensuring data integrity and reducing accidental loss of data through automation.

❌ Incorrect Options:

🟥 A. Agent Delete Record Action permission is not associated to the user.
This isn’t relevant because there’s no “Delete Record Action” permission specific to the Agent. The issue isn’t a missing permission—it’s a lack of feature support. Adding or modifying permissions won’t resolve the problem.

🟥 B. Agent does not have the permission to delete the user's records.
While permission settings can restrict actions, in this case, the Agent wouldn’t be able to delete any records at all, regardless of permissions, since deletion isn’t part of its functional scope.

📘 Reference:
Salesforce Help: Supported Agent Actions

Explanation:

Summary:
This scenario blends service verification with sales actions like quoting and escalation for high-value upsells, highlighting the need for an agent tuned to revenue growth. In Agentforce, selecting the right agent type streamlines workflows across channels, ensuring secure handling of customer data while automating routine tasks and handing off complex deals, ultimately shortening cycles and lifting conversions.

Correct Option:

✅ B. Sales Agent to handle the upsell and large-deal escalation.
Perfect for this flow, as it verifies entitlements via CRM, generates quotes with prorated calcs, and triggers AE escalations for deals over $25K—core upsell mechanics. Agentforce Sales Agents, like SDRs, nurture opportunities 24/7 using trusted data, reducing manual handoffs and boosting win rates by focusing reps on big closes.

Incorrect Options:

❌ A. Service Agent to resolve the case end-to-end and create a new opportunity for the sales team.
Service Agents excel at resolving issues like tickets but aren't optimized for proactive quoting or value-based escalations; they'd likely defer the opportunity creation, delaying upsells. In Agentforce, this shifts sales burden back to humans, missing the autonomous revenue push needed here.

❌ C. Employee Agent to orchestrate internal logistics and finance.
These are internal tools for staff tasks like HR or IT, not customer-facing upsells—they lack built-in quoting or escalation for external interactions. Agentforce Employee Agents support workflows behind the scenes but wouldn't directly engage verified customers, risking compliance gaps in entitlement checks.

Reference:
Explore Agentforce: Your Guide to Autonomous Agents (Salesforce Trailhead)
What is a Sales Agent? Types, Roles, & More

Explanation:

Summary:
In Agentforce implementations, handling ambiguous customer queries like "package" requires a retrieval strategy that merges exact matching for reliability with intent understanding for relevance. This approach ensures accurate results in dynamic scenarios, such as resort bookings, by leveraging Salesforce Data Cloud's capabilities to ground AI responses in trusted data, reducing errors and boosting user satisfaction without over-relying on one method.

Correct Option:

✅ A. Use hybrid search, which combines keyword matching for precision with semantic embeddings for context.
This strategy shines in ambiguous situations by fusing keyword search's exact-match precision—ideal for specific terms like "baggage"—with semantic embeddings that grasp broader context, such as vacation intent. In Agentforce, it powers Retrieval-Augmented Generation (RAG) for more accurate, context-aware responses, as seen in Data Cloud integrations. Result: Fewer misinterpretations and higher retrieval quality for customer interactions.

Incorrect Options:

❌ B. Use semantic search only, which captures intent but may struggle with ambiguous terms when no context is provided.
While great for understanding overall meaning, pure semantic search falters on vague queries lacking clues, potentially pulling unrelated content like general "packages" instead of baggage specifics. In Agentforce, it misses the precision needed for domain jargon, leading to less reliable RAG outputs and requiring extra user clarification, which slows service.

❌ C. Use keyword search only, which prioritizes exact term matching but risks missing contextual meaning.
This method locks onto literal words, nailing precision for "package" but ignoring nuances—like vacation vs. luggage—resulting in irrelevant or incomplete results. For Agentforce, it limits AI's ability to disambiguate without semantic layers, often forcing escalations and frustrating users in natural-language chats.

Reference:
How Data Cloud Hybrid Search Combines Keyword and Vector Retrieval to Elevate the Search Experience
Agentforce and RAG: Best Practices for Better Agents