Salesforce · AI-201
Salesforce Certified Agentforce Specialist (AI-201) Practice Tests
Validates expertise in managing and optimizing Agentforce, including building custom AI agents, configuring Prompt Builder, and leveraging Data Cloud to drive AI-powered automation within the Salesforce platform.
Exam Details
Practice Questions
600
≈ 10 practice exams
Duration
105 minutes
Passing Score
73%
Difficulty
AssociateLast Updated
May 2026
Topics Covered
Agentforce Agent Configuration and CustomizationPrompt Builder and Prompt Template ManagementAgentforce Testing Center and ValidationAgentforce Data Library and Data Cloud RetrieversAI Actions, Integrations, and Custom SolutionsSecurity, Monitoring, and Responsible AI UsageStakeholder Engagement and Use Case Identification
Salesforce Certified Agentforce Specialist (AI-201) Practice Exam Preparation
Use this AI-201 practice exam to prepare for Salesforce Certified Agentforce Specialist (AI-201) with realistic questions, detailed explanations, and focused study modes. The practice bank includes 600 questions for Salesforce AI-201, so you can review the exam steadily instead of relying on one long cram session.
As you practice, pay extra attention to recurring topics such as Agentforce Agent Configuration and Customization, Prompt Builder and Prompt Template Management, Agentforce Testing Center and Validation, Agentforce Data Library and Data Cloud Retrievers, and AI Actions, Integrations, and Custom Solutions. Start with short sessions to identify weak areas, then move into timed quizzes once your accuracy is consistent.
The explanations are especially useful when you want to connect exam wording to the responsibilities and scenarios described in the official certification guidance. Use the free preview first, then unlock the full question bank when you are ready to build a complete study routine.
Exam Domain Breakdown
AI Agents35%
Prompt Engineering20%
Data Cloud for Agentforce20%
Development Lifecycle20%
Multi-Agent Interoperability5%
Exam Overview
The Salesforce Certified Agentforce Specialist (AI-201) validates a professional's ability to build, configure, and optimize AI-powered agents using Salesforce's Agentforce platform. Introduced on March 3, 2025, this certification reflects Salesforce's strategic push to embed autonomous AI agents across its ecosystem, covering core competencies such as agent topic and action configuration, the Agentforce reasoning engine, Prompt Builder template creation, and Data Cloud retrieval-augmented generation (RAG) grounding techniques. Candidates are tested on how to architect agents that handle real business workflows across channels including digital experience sites, email, and Slack.
The certification also encompasses the full agent development lifecycle — from initial design through sandbox testing in the Agentforce Testing Center to production deployment — as well as emerging capabilities like multi-agent interoperability via the Model Context Protocol (MCP) and the Agent API. It supersedes the earlier Salesforce Certified AI Specialist credential, expanding its scope to address the Agentforce product suite that spans Service, Sales, and Employee agent types.
Who Should Take This Exam
This certification is designed for Salesforce professionals who are responsible for implementing or managing Agentforce within their organizations. Typical roles include Salesforce Administrators, Platform App Builders, Functional Consultants, and Solution Architects who need to design and configure AI agents without necessarily writing code. It is also relevant for AI/Automation Specialists and pre-sales engineers who need to demonstrate hands-on proficiency with Salesforce's AI tooling.
Candidates are expected to have practical, hands-on experience working with Agentforce in a real or developer org environment. The exam does not require a software engineering background, but a working understanding of Salesforce configuration fundamentals — such as flows, objects, permissions, and data models — is essential for success.
Prerequisites
There are no formal prerequisites required to sit for the AI-201 exam. However, Salesforce recommends that candidates hold or have working knowledge equivalent to the Salesforce Certified Associate or Salesforce Certified Administrator credential before attempting this exam, as familiarity with core platform concepts (objects, permissions, automation) is assumed throughout.
Practical experience with Agentforce is strongly advised. Candidates should have hands-on time configuring agents, building prompt templates in Prompt Builder, and working with Data Cloud data sources in a Developer Edition or sandbox org. Completing the official Trailhead learning path 'Become an Agentblazer' and the 'Cert Prep: Salesforce Agentforce Specialist' Trailhead module before sitting the exam is highly recommended.
Exam Format
The AI-201 exam consists of 60 scored multiple-choice and multiple-select questions, with an additional 5 unscored pilot questions embedded throughout, bringing the total number of items presented to approximately 65. Candidates are given 105 minutes to complete the exam. The minimum passing score is 73%. The registration fee is $200 USD, and retakes cost $100 USD.
The exam is available via online proctored delivery or at an authorized Pearson VUE testing center. No physical materials or external resources are permitted during the exam. Like all Salesforce certifications, the AI-201 requires annual maintenance through completion of release-specific Trailhead modules to remain active.
Skills Measured
- 1.AI Agents (35%): Configuring agent behavior using the reasoning engine, managing deterministic controls with filters and variables, selecting and configuring standard and custom topics and actions for Service, Sales, and Employee agent types, managing Agentforce security including the Agent User concept, and connecting agents to channels such as digital experiences, email, and Slack.
- 2.Prompt Engineering (20%): Using Prompt Builder to create and manage prompt templates including Field Generation and Flex template types, assigning and managing Prompt Builder user roles and permissions, selecting appropriate grounding techniques (merge fields, RAG), and applying best practices to reduce hallucinations and improve prompt clarity and effectiveness.
- 3.Data Cloud for Agentforce (20%): Understanding Agentforce Data Library types, improving agent response accuracy using unstructured data through chunking and vector indexing, implementing retrievers in Data Cloud to ground agent responses, and selecting among keyword, vector, and hybrid search strategies based on use-case requirements.
- 4.Development Lifecycle (20%): Testing agents and validating behavior using the Agentforce Testing Center, identifying the steps required to deploy agents from sandbox to production environments, and describing the process for monitoring agent adoption and performance post-deployment.
- 5.Multi-Agent Interoperability (5%): Understanding the Model Context Protocol (MCP) and its role in enabling agent-to-agent communication, determining when and how to use agent-to-agent protocols, and identifying the correct steps for using the Agent API to integrate Agentforce with external systems.
Study Tips
- Spin up a free Agentforce Developer Edition org and build at least one end-to-end agent from scratch — configure topics, actions, and channels — before sitting the exam, as the test heavily emphasizes hands-on configuration knowledge.
- Complete the official 'Cert Prep: Salesforce Agentforce Specialist' Trailhead module, which maps directly to exam domains and contains unit-by-unit review content aligned to the five tested areas.
- Work through the 'Become an Agentblazer' Trailhead learning path in full; this is Salesforce's official preparation trail and covers Prompt Builder, Data Cloud retrievers, and the Testing Center in structured sequence.
- Pay particular attention to the AI Agents domain (35% weight) — specifically the distinction between Service, Sales, and Employee agent types, how the reasoning engine selects topics and actions, and the Agent User security model, as these are the highest-yield areas.
- Practice building multiple prompt template types in Prompt Builder (especially Flex templates) and experiment with different grounding techniques (merge fields vs. RAG) to develop intuition for when each approach is appropriate — this covers 20% of the exam.
- Use the Agentforce Testing Center in your dev org to test agent conversation flows and understand how to interpret test results; the Development Lifecycle domain (20%) includes testing, deployment steps from sandbox to production, and adoption monitoring.
- Review Salesforce's documentation on the Model Context Protocol (MCP) and the Agent API to ensure coverage of the Multi-Agent Interoperability domain, even though it represents only 5% of the exam — these topics are frequently misunderstood and easy to lose points on.
Career Benefits
Earning the Salesforce Certified Agentforce Specialist positions professionals at the forefront of enterprise AI adoption, as Agentforce is Salesforce's flagship AI product and a core part of its go-to-market strategy. Roles that directly benefit from this certification include Salesforce AI Specialist, CRM AI Consultant, Salesforce Administrator (Senior), and Salesforce Solution Architect, all of which are experiencing strong demand as organizations accelerate AI agent deployments within their CRM platforms.
As one of the first certifications dedicated specifically to agentic AI on the Salesforce platform, the AI-201 differentiates holders in a market where Agentforce expertise is scarce relative to demand. It complements broader Salesforce credentials (Admin, Platform App Builder, Architect) and is increasingly listed as a preferred qualification in Salesforce implementation and consulting job postings. Notably, Salesforce offered this exam free of charge through December 31, 2025 as part of its Agentblazer initiative, signaling strong organizational investment in building a certified practitioner community.
Sample Questions
5 sample questions with answers and explanations. The full bank has 600 questions, enough for 10 full-length practice exams.
Preview — answers shown1. A support agent frequently routes “I moved and need to change where my order ships” to the General Account Updates topic instead of the Order Shipping Changes topic. The Order Shipping Changes instructions are detailed and correct, including rules for shipped versus unshipped orders. Which configuration should be changed first? (Select one!)
AAdd a confirmation requirement to the shipping update action so routing decisions become more conservative.
BRevise the Order Shipping Changes classification description to include address-change, delivery-address, ship-to, and moved-address user intents, and clarify that account profile updates are out of scope.
CAdd more steps to the Order Shipping Changes instructions that explain how to update an address after the topic is selected.
DChange the action output description for the shipping update Flow so the topic classifier has more context.
Explanation
The symptom is wrong topic selection. The first fix is to improve the routing metadata: classification description and boundaries for the relevant topic. Detailed instructions and action settings matter only after the correct topic is active.
2. A telecom company has separate topics for Plan Changes, Billing Disputes, and Device Troubleshooting. Users who say 'My bill went up after changing my plan' are handled inconsistently. What topic-design change best addresses the ambiguity? (Select one!)
AAdd the same billing, plan-change, and troubleshooting actions to every topic so routing has less impact on available actions.
BMerge all three areas into a broad Account Help topic so the agent does not need to distinguish user intent.
CClarify the in-scope and out-of-scope boundaries for Plan Changes and Billing Disputes, including examples of plan-related billing complaints and where they should route.
DDeactivate Billing Disputes so all mixed plan-and-bill requests start in Plan Changes.
Explanation
Mixed user requests require clearer topic boundaries and examples so the agent can choose the most appropriate topic. Deactivating topics, collapsing unrelated domains, or mapping the same actions everywhere reduces control and can worsen ambiguity.
3. A user asks an agent, “Can you find my open case, check whether the replacement part shipped, and tell me what happens next?” Which sequence best represents the Atlas reasoning pattern? (Select one!)
ARoute the utterance by exact keywords, ignore previous context, and require the user to restart for each subtask.
BClassify the topic, select relevant actions, fill required inputs from context or by asking, execute and observe action results, then generate a grounded response.
CGenerate a final response from the LLM first, then search for actions only if the user challenges the answer.
DRun every action mapped to every topic, then choose the topic whose output is the longest.
Explanation
The reasoning loop involves understanding the request, selecting a topic and actions, collecting inputs, executing actions, observing results, and producing a grounded response. This is more than keyword matching or single-step text generation.
4. A healthcare organization is configuring an agent that can answer appointment and office-hour questions but must transfer users who ask for medical advice. Which guardrail design is strongest? (Select one!)
AGive the agent broad clinical data access so it can decide whether advice is appropriate
BRely on a friendly agent-level instruction that says the agent should be careful with medical questions
CCombine narrow topic scope, explicit medical-advice out-of-scope rules, a defined escalation path, and permission limits on what the agent can access or change
DUse only toxicity detection because medical-advice requests are unsafe content
Explanation
Sensitive use cases require layered guardrails: topic boundaries, escalation behavior, permissions, and Trust Layer protections. A soft instruction alone is weaker than combining configuration scope, access control, and a clear handoff path.
5. An agent uses a prompt template to draft a concise renewal-risk explanation for an account manager. Testers report that the draft contains plausible but unsupported reasons that are not present in CRM data or approved knowledge. Which two changes best address the issue? (Select two!)
Multiple correct answersAGround the prompt template with relevant account, opportunity, or case data that supports the renewal-risk explanation.
BConfigure retrieval from approved knowledge or indexed content when the explanation depends on policy or product information.
CMove the renewal-risk logic into the topic classification description so the topic router can add missing business facts.
DReplace the prompt template with a delete-record action and require confirmation before the action runs.
ERemove all input variables from the template so the LLM relies only on general model knowledge.
Explanation
Grounding is the key control for reducing unsupported generated content. Record data and approved retrieval sources give the prompt template trusted context. Topic classification descriptions route requests; they do not supply execution-time facts. Removing grounding increases hallucination risk.
More Salesforce Practice Exams
Salesforce Certified Advanced Administrator (CRT-211)
CRT-211 · 600 questions
Salesforce Certified Experience Cloud Consultant (EX-Con-101)
EX-Con-101 · 600 questions
Salesforce Certified Marketing Cloud Account Engagement Specialist (MC-201)
MC-201 · 600 questions
Salesforce Certified Marketing Cloud Email Specialist (MC-202)
MC-202 · 600 questions
Salesforce Certified Marketing Cloud Engagement Consultant (MCE-Con-201)
MCE-Con-201 · 600 questions
Salesforce Certified Platform Administrator
SCA · 600 questions
$17.99
One-time access to this exam
600 questions (10 practice exams' worth)
Unlimited timed exam simulations
Or $15/mo for all 253 exams
Detailed explanations
Free preview stays available