A free Agentforce-enabled Developer Edition org (sign up from Trailhead) for every lab below
Week 1 · 8 h
Prompt Builder and the Einstein Trust Layer
Decide when Prompt Builder fits a requirement and choose a template type (Sales Email, Field Generation, Record Summary, Flex)
Ground templates with merge fields, related lists, flows, Apex and Data 360 retrievers
Create, activate, version and invoke templates, and control who can use them with permission sets
Explain Trust Layer features (data masking, zero data retention, toxicity scoring, audit trail) and how to block specific models
Hands-on: Build a Field Generation template that writes a case summary into a custom field, ground it with a related list and a flow, activate it, and test it from the record page as a user with and without the permission set.
Explain agent building blocks and Agent Script, and how hybrid reasoning works across Canvas and Script views
Make behaviour deterministic with filters, variables and template expressions
Configure standard and custom topics and actions (flow, Apex, prompt template), and write clear topic instructions and classification descriptions
Choose between an Employee agent and a Service agent, and know when to use the Agent API
Connect agents to channels (Experience Cloud, email, voice, Slack) and explain the running user's security context for actions
Hands-on: Build a Service agent with one custom topic and two actions (a flow that looks up an order and a prompt-template action). Add a variable-based filter so the refund action only runs for verified customers, then deploy it to an Experience Cloud site.
Write test cases in Testing Center, run batch tests and read the evaluation results
Move agents and prompt templates from sandbox to production, including the activation and permission steps
Monitor agents with analytics and optimisation insights, and explain the role of MCP and A2A in multi-agent architectures
Hands-on: Write 15 test utterances in Testing Center for your agent, including off-topic ones. Fix the topic instructions until every test passes, and record what you changed.
Week 4 · 6 h
Practice and review
Take timed practice tests (60 questions in 105 minutes) and aim for at least 80%
Re-study AI Agents (35%) first, then Prompt Engineering and Data 360 (20% each)
Check which Salesforce release the current exam guide maps to, and review release notes for anything newer
You have built, tested in Testing Center and deployed a working agent in your own org
You can pick the right prompt template type and grounding method for any scenario in under 15 seconds
You can explain Agent Script, hybrid reasoning and the running user's security context without notes
You score 80% or higher on two different timed practice tests
Exam tips
The exam is 60 scored questions plus up to 5 unscored, 105 minutes, with a 72% pass mark. Questions currently align to the Spring '26 release.
Agentforce changes every release and Salesforce retires older content, so make sure your study material covers Agent Script and Data 360 rather than old Copilot or Data Cloud wording.
The credential needs an annual Trailhead maintenance module, and a retake costs $100.
Start with the provider's official exam guide and free training, then use practice exams to find your weak domains. We never link to exam dumps, which break exam agreements and can get you decertified.