Agentforce Specialist study guide (AGENTFORCE-SPEC)

The Salesforce Certified Agentforce Specialist exam tests whether you can design and implement prompt templates and AI agents on the Salesforce Platform, grounded in Data 360 and governed by the Trust Layer. It is aimed at administrators, developers and architects with about a year of Salesforce configuration experience. You answer 60 scored multiple-choice questions, plus up to five unscored ones, in 105 minutes and need 72% to pass.

AGENTFORCE-SPEC at a glance

Details as of September 2026 — confirm on the official exam page before booking.

Who this exam is for

Salesforce's exam guide describes a candidate with around one year of experience configuring the Salesforce Platform and standard objects, including Data 360, and hands-on experience building agents in Agent Builder and prompt templates in Prompt Builder. They should also know Testing Center and the usual sandbox-to-production deployment patterns. There is no prerequisite, though Salesforce recommends the Platform Administrator and Platform App Builder certifications.

The guide is also clear about what you do not need: LLM fine-tuning, Apex or Python coding, transformer internals, ROI calculations, or expertise in products such as Marketing Cloud, MuleSoft, Tableau or Slack. Slack does appear as a channel an agent can be connected to, but not as a product to master. The exam is closed book, and the certification requires an annual maintenance module on Trailhead.

What the exam covers

Questions align to the Spring '26 release. Salesforce renames Agentforce features between releases (topics, for example, are now documented as subagents, and Data Cloud became Data 360), so study the Spring '26 behaviour and treat the exam guide's vocabulary as the vocabulary of the exam.

Prompt Engineering (20%)

This domain covers when Prompt Builder is the right tool, which prompt template type fits (field generation, flex and the others), how templates are created, activated and run, and who can access them. Grounding is central: choosing the right source for a scenario, such as record data, a flow or a Data 360 retriever. It also covers prompt-writing practice, the Trust Layer's security and privacy features, and preventing access to specific models.

Data 360 Fundamentals (20%)

A fifth of the exam from only two objectives, so each is examined in depth. You need to understand Agentforce Data Library and what it sets up for you, and the retrieval chain underneath: how content is chunked, indexed and exposed through retrievers to agents and prompt templates. Expect scenarios where the answer depends on which stage of that chain is misconfigured.

AI Agents (35%)

The largest domain. It covers how an agent reasons and its building blocks in Agent Script, hybrid reasoning in Canvas and Script View, and making behaviour deterministic with filters, variables and template expressions. You choose and configure standard and custom topics and actions, connect agents to channels (digital experience, email, voice, Slack), decide between an Employee and a Service agent, and judge when Agent API is appropriate. The security context an agent runs under, and how it affects what actions can do, is a frequent source of scenario questions.

Testing, Deployment, and Maintenance (10%)

This domain covers testing agents in Testing Center and how its evaluations work, and what to consider when moving agents and prompt templates from sandbox to production.

Governance and Observability (10%)

Managing and monitoring agents in production, and using agent analytics to find and fix weak behaviour.

Multi-Agent Orchestration (5%)

The smallest domain: deciding whether a multi-agent architecture is justified for scale and control, and explaining the purpose of open protocols such as MCP and A2A.

A 4-week study plan

This plan assumes six to eight hours a week in a Developer Edition or sandbox org with Agentforce enabled. AI Agents is 35%, so it gets most of two weeks.

Week 1: Prompt Engineering (20%). Build one template of each common type. Ground one with record fields, one with a flow and one with a retriever, and compare the resolved prompt in preview. Review the Trust Layer features and how model access is restricted. Finish with a mixed paper.

Week 2: Data 360 (20%) and the start of agents. Set up an Agentforce Data Library on a small set of documents and trace what it creates. Learn chunking, indexing and retrievers well enough to diagnose a poor answer. Then build a first agent and read its Agent Script.

Week 3: AI Agents (35%). Add custom topics and actions, use variables and filters to force a deterministic step, and test what happens when the running user lacks a permission. Connect the agent to a channel. Write down, for five scenarios, whether an Employee or Service agent fits and why.

Week 4: Testing, governance, multi-agent (25% combined) and rehearsal. Run a Testing Center batch, then walk through deploying an agent and a template to production. Review analytics and the MCP and A2A basics. Sit a full-length mock in 105 minutes, drill weak topics, then sit a second. See [how to use practice exams effectively](/blog/how-to-use-practice-exams-effectively) for review technique.

Where candidates lose marks

• Underweighting Data 360. Two objectives look light on paper, but together they are worth as much as all of Prompt Engineering. Candidates who skim chunking, indexing and retrievers lose a fifth of the exam's marks cheaply.
• Security context. Scenarios where an action fails often turn on which user the agent runs as and what that user can access, not on the action's configuration. Check permissions first.
• Studying a later release. Current documentation describes features added after Spring '26. If an option describes something you saw only in a newer release note, be cautious.
• Employee versus Service agent. The deciding facts are who the user is and whether they are authenticated inside Salesforce. Read the scenario for the audience before the requirement.
• Grounding choice. Several grounding methods can technically work. The better answer is the one that fits where the data lives and how current it must be.
• Deployment detail. Moving templates and agents from sandbox to production brings metadata and dependencies with them. Know what has to travel together, rather than assuming a single deployment moves everything.

Practising for AGENTFORCE-SPEC on CertifyAIx

CertifyAIx has 670 original Agentforce Specialist practice questions, written against the Spring '26 exam guide and weighted by its six domains, each with explanations. Salesforce publishes no in-scope feature list for this exam, so our coverage follows the guide's objectives. Read [how CertifyAIx practice questions are written](/blog/how-certifyaix-practice-questions-are-written) for the method.

• 3 full-length mock exams, timed at the real exam's 105 minutes
• 29 mixed domain papers and 35 topic papers of up to 25 questions each (60 minutes per 25)
• Progress tracked by domain and objective

Start with the [AGENTFORCE-SPEC practice questions](/certifications/salesforce-agentforce-specialist).

Frequently asked questions

How much experience do I need?

Salesforce describes a typical candidate as having about a year of platform configuration experience, including Data 360, with hands-on agent and prompt template building. Coding is not required.

How long should I prepare?

An admin who already builds agents might need three to four weeks. Someone new to Agentforce or Data 360 should allow six weeks or more, because both need hands-on practice in an org rather than reading alone.

Are CertifyAIx questions real Salesforce exam questions?

No. Every question is original and written from the public exam guide. Salesforce requires candidates to accept its certification agreement and code of conduct; our article on [why exam dumps put your certification at risk](/blog/why-exam-dumps-put-your-certification-at-risk) explains the stakes.

What happens if I fail?

You can retake the exam; Salesforce lists the retake fee at USD 100 plus applicable taxes.

How do I keep the certification current?

Complete the Agentforce Specialist maintenance module on Trailhead once a year. If you miss the due date, the certification expires.

CertifyAIx is independent and not affiliated with Salesforce. Salesforce Certified Agentforce Specialist is a trademark of its owner. All CertifyAIx practice questions are original.