Advanced track for implementing AI Agents in the Salesforce environment
Salesforce Agentforce Course
Advanced training for implementing AI Agents in the Salesforce environment
Your next step in Salesforce: From business process design to building a working, secure, and tested AI agent. The course is intended for those with a suitable Salesforce background or graduates of HPI's Salesforce Implementer course.
120 academic hours
30 sessions
10 labs
Practical capstone project
What is Agentforce?
Agentforce allows an agent to understand a natural language request, identify the user's context and identity, use only Salesforce data and permitted knowledge sources, select an authorized action, and complete a business process. At sensitive points, Flow, Agent Script, human approval, and transfer to a representative are integrated. The decision rule is simple: a known and fixed path is suitable for Flow, and when understanding intent, context, and choosing between tools is required - Agentforce may be suitable.
Get the full syllabus and check suitability for the track
Leave your details and we will get back to you with the full Agentforce course syllabus and a short conversation to check if your Salesforce background is suitable for the track.
Professional Advancement in Salesforce
Add to your existing platform knowledge the ability to design, build, and deploy AI Agents – not just configure automations.
Building a Practical AI Solution
10 labs, 75 academic hours of practice, and a capstone project simulating an organizational implementation in a controlled environment.
For individuals and teams in organizations
A public track for personal practice and a portfolio, or organizational training around use cases, security, governance, and value measurement.
Why study Agentforce at HPI
From business need to an AI agent working in the Salesforce environment
This track is for those who already work with Salesforce and want to know how to design, build, secure, test, and implement AI agents. It does not teach Salesforce from scratch.
Start from the business process
Learn to perform Discovery, define a Use Case with clear boundaries, prioritize by KPI, cost, risk, and ROI, and decide when Agentforce is suitable and when Flow or another solution is preferable.
Build with the platform's real tools
Work in Agentforce Builder, Agent Script, Prompt Builder, and Flow Builder, and connect Subagents, Actions, and Instructions into a complete workflow.
Security, governance, and human approval
Implement Least Privilege, Guardrails, controls against Prompt Injection and information leakage, and integrate human approval and handoff to an agent at sensitive points.
Test and deliver like in an enterprise project
Build Evaluation Sets, analyze failures, perform Red Teaming, and submit an implementation file with an architecture document, Demo, and operational plan.
The practical advantage
62.5% of the track is practical work: 75 practice hours, 10 labs, and a final project
10 labs with a tested output
Each lab practices one clear task and ends with a verifiable output: a successful scenario, a failure scenario, and permissions testing. Practice is done with synthetic or anonymous data.
Grounding on authorized data and knowledge sources
Connect Salesforce Knowledge, Data Library, and Data 360 as available, and check if the agent retrieves the correct information and only the information it is authorized to access.
Integration with Flow, Apex, and APIs
Build Agent Actions on top of Flow, connect actions to external systems with Apex, APIs, and MuleSoft, and practice authentication and error handling with a Mock API.
Deployment, monitoring, and Kill Switch
Work with Agentforce DX, Salesforce CLI, and Git, and plan for monitoring, Rollback, Kill Switch, and a user adoption plan.
Before Deciding
The Agentforce track requires the right starting point. Let's really check it out
We have gathered the important questions about the required background, what an Agentforce implementer actually does, the extent of practice, the code, security, and the deliverables - so you can understand if this is the right professional stage for you or your team.
120 academic hours
90 clock hours
30 sessions × 4 hours
12 modules
10 labs
75 practice hours
Production-like final project
01
Is my Salesforce background sufficient for the course?
The track does not teach Salesforce from scratch. It is intended for those who have already worked on the platform and want to learn how to build and implement Agentforce.
The course is intended for those with suitable Salesforce backgrounds or graduates of HPI's Salesforce Implementer course. Relevant background includes familiarity with the Salesforce data structure - Objects, Fields, and Relationships, basic work in Salesforce Flow, managing Users, Permission Sets, CRUD, FLS, and Sharing, and familiarity with Sales Cloud or Service Cloud. Technical English is also required for reading interfaces and documentation.
The target audience includes Salesforce Administrators, implementers and consultants, Business Analysts, information systems and CRM professionals, Flow, automation, development and integration professionals, as well as project managers, Solution Architects, and transformation leaders. For absolute beginners, an HPI Salesforce Implementer course or a separate Pre-Course of 40-60 hours is required first; these hours are not part of the 120 course hours.
They turn a business need into an Agent that understands a request, operates within defined boundaries, and completes a process - including security, testing, and delivery.
The work begins by clarifying the need: who are the users, what is the process, what is measured, and what is the risk. Afterward, a Use Case is defined, the Agent is built in Agentforce Builder with Subagents, Instructions, and Actions, connected to authorized data and knowledge sources, and Flow, Apex, or APIs are integrated for operations requiring additional systems.
Finally, test sets are built, failures are analyzed, Guardrails and human approval are implemented, and the solution is transferred between environments with monitoring, Rollback, and an adoption plan. Course graduates can explore roles such as Product Owner, Business Analyst, Consultant, Salesforce Admin, or Agentforce Implementer with AI specialization, depending on experience and employer requirements.
03
What is the difference between an AI Agent, a Chatbot, and regular automation?
Automation executes a fixed path, a chatbot answers according to a script, and an agent identifies intent and context and chooses between authorized tools.
When the path is known, fixed, and pre-definable – Flow is the right solution, and it is also more predictable and cheaper to maintain. A classic script-based chatbot reaches its limitation as soon as the user phrases the request differently. An Agentforce agent identifies a goal, user, and identity, selects appropriate Subagents and Actions, retrieves authorized information or performs an action, and escalates to a human when needed.
In the first module, we learn this decision rule and practice it on a real scenario: we analyze a case and choose between Agentforce, Flow, or another solution, including the limitations of AI models. The output is a concept map and a short decision document.
04
How much do we really practice and what do we build?
75 out of 120 academic hours are practice – 62.5% of the course – alongside 27 hours of theory and 18 hours of demonstrations.
The course includes 10 practical labs. Each lab practices one clear task and concludes with a verifiable deliverable, including: Use Case Canvas and Scorecard, Data Readiness and Security Report, Prompt Template and Test Suite, Agent with Subagents and Actions, two Agent Actions based on Flow, integration with Mock API, connected Knowledge source and retrieval quality test, risk and control mapping and cost, Evaluation report and Go/No-Go decision.
In each lab, we test a valid scenario, a failure scenario, and permissions. Practice is done with synthetic or anonymous data, and an alternative path is defined for Beta capabilities.
05
Do you need to know how to code?
Most building is done with declarative tools, but the course also includes an Apex, APIs, and Integrations module – this is not a no-code course.
The majority of the work is done in Agentforce Builder, Prompt Builder, Agent Script, and Flow Builder, meaning with declarative tools. Concurrently, Module 07 covers Apex, APIs, MuleSoft, and integrations, including connecting actions to systems, authentication, error handling, and practice with a Mock API, and the final project requires integration or code.
Therefore, the course should not be presented as a no-code course. On the other hand, the admission requirements are those listed in the syllabus: Salesforce background and technical English. Prior development experience is helpful but not defined as a prerequisite, and logical thinking and willingness to work with technical interfaces are required.
06
How do you check that the agent is operating correctly and securely?
Two modules are dedicated to Trust, Security, and Governance, and to Testing, Evaluation, Red Teaming, and Observability.
On the security side, we work according to Least Privilege, define Guardrails, check CRUD, FLS, and Sharing from the perspective of different users, deal with Prompt Injection, and prevent information leakage. At sensitive points, human approval is integrated before performing an action. Concurrently, we examine consumption, cost, and ROI.
On the testing side, we build test scenarios, run Evaluation Sets, read Traces, analyze failures and improve the agent, and perform Red Teaming. The output is an Evaluation report and a Go/No-Go decision. The final project requires at least 30 Evaluation cases and security tests.
07
What will I present at the end of the course and how does it support my professional development?
Production-like final project: a working agent, Flow, integration or code, authorized knowledge sources, tests, security, and a handover package.
The final project simulates an organizational implementation and is submitted as a complete package: a working agent and professional Demo, at least one Flow, integration or code, authorized knowledge sources, at least 30 Evaluation cases, security tests, architecture document and technical documentation, as well as an implementation package including tests, risks, and an operational plan.
The project is carried out in a controlled environment and with synthetic data only – not on actual customer data – and does not constitute a promise of deployment to a customer, work experience, or a commitment to placement. It does allow you to explain in an interview or within the organization how you planned, secured, tested, and delivered an AI solution.
08
Is the training suitable for me personally or for a team in an organization?
Both tracks exist: a public track for individuals and organizational training for teams.
For individuals: Deepening implementation capabilities, practical training in 10 labs, a project for the portfolio, and professional preparation for the knowledge areas of the certification. Suitability is checked in a short conversation before registration, based on existing Salesforce background.
For organizations: Training teams around the company's use cases, business processes, security, governance, implementation, and value measurement. If you are considering training for a team, please indicate this on the form and we will get back to you with an appropriate contact path.
Want to check if your Salesforce background is suitable for the Agentforce track?
Tell us what you do today in Salesforce - data model, Flow, permissions, Sales Cloud or Service Cloud. We will get back to you for a short call, send the full syllabus, and help you understand if this is the right starting point, for an individual or a team within an organization.
No obligation · Short call · Personal response from the HPI team
Want to check suitability and receive Agentforce course details?
Leave your details and we will get back to you with the full syllabus, an suitability assessment based on your Salesforce background, and updated registration details and dates. If you are considering training for a team in your organization, we will build a track with you around your use cases.
No obligation · Short call · Personal response from the HPI team
Curriculum
12 modules, 120 academic hours, and 10 labs - from Use Case to Production-like
The track progresses from Agentforce and Discovery fundamentals to data readiness and security, Prompt Builder, building in Agentforce Builder, automations, integrations, Knowledge and RAG, governance, testing, deployment, and a final project.
Understand what Agentforce is, how it works, and when it is appropriate to use it. This module establishes the decision framework between an AI agent and existing automation.
The difference between Agent, AI, Flow, Prompt, and Chatbot
Agentforce components and their operation
Limitations of AI models
Choosing the right solution for the scenario
Lab: Scenario analysis and choice between Agentforce, Flow or another solution
Deliverable: Concept map and a short decision document for choosing the appropriate solution.
Transform a business need into a clear, measurable, and implementable use case, with boundaries, risks, and success metrics.
Discovery: Clarifying the need, users, and business process
Defining the Use Case and solution boundaries
KPIs and success measurement
Cost, risk, and ROI
Initial solution architecture
Lab: Scorecard and comparison of use cases
Deliverable: Use Case Canvas and Scorecard for comparing scenarios.
Verify if the Salesforce environment is ready to operate an agent securely, in terms of data, Knowledge, and identities.
Data structure and data quality
Available knowledge sources
User identity and Permission Sets
CRUD, FLS and Sharing tests
Lab: Testing data and permissions from the perspective of different users
Outcome: Data and security readiness report.
Build clear prompts for the AI and provide it with correct and authorized information, so that the answer is based on a real source.
Prompt Builder: Building prompts and templates
Context Engineering: Context planning
Grounding: Basing the answer on data and authorized sources
Reusable Prompt templates
Quality assurance: Success and failure scenarios
Lab: Building a Prompt Template and testing it
Outcome: Prompt template and a set of quality tests.
The central module of building: define an Agent, break it down into Subagents, connect Actions, and write Instructions and routing paths.
Agentforce Builder
Subagents and division of responsibility
Actions and authorized operations
Instructions and writing work instructions
Agent Script
Request routing and path selection
Multi-Agent and transfer scenarios
Transfer to a human agent
Lab: Agent with Subagents and Actions
Outcome: A working Agent with Subagents, Actions and defined routing paths.
Connecting the Agent to declarative processes: retrieval, controlled update, validation, permissions, and error handling.
Building a Flow for information retrieval
Building a Flow for controlled updates
Input, output, and validation checks
Permissions for each action
Error handling
User approval before sensitive action
Lab: Two Agent Actions based on Flow
Output: Two Flow-based Agent Actions, including validation and human approval.
Connecting Agent actions to systems outside Salesforce, with authentication, failure handling, and practice with a Mock API.
Connecting Actions to external systems
Apex in the context of agent actions
APIs and integration
MuleSoft, subject to availability
Authentication and permissions between systems
Handling failures and response times
Lab: Integration with a Mock API
Output: Integrated action with an external system, including failure handling.
Grounding the Agent on authorized knowledge sources and verifying that it actually retrieves the correct information.
Preparing Salesforce Knowledge and documents for Agent use
RAG: Retrieval Augmented Generation
Agentforce Data Library
Data 360, subject to availability
Retrieval quality testing
Lab: Connected Knowledge Base and Retrieval Test
Deliverable: Connected Knowledge Source and Retrieval Quality Test.
Defining the agent's boundaries and surrounding controls, and examining the cost and value of operation.
Least Privilege
Guardrails and Usage Policy
Prompt Injection and Preventative Controls
Preventing Information Leakage
Human Approval at Sensitive Points
Consumption, Cost, and ROI
Lab: Risk Mapping, Controls, and Cost
Deliverable: Risk and control mapping with cost and value considerations.
Testing the agent against predefined scenarios, analyzing failures, and deciding if the solution is mature enough to proceed.
Building Test Scenarios
Evaluation Sets
Red Teaming
Traces and Agent Behavior Monitoring
Failure Analysis and Improvement
Go/No-Go Decision
Lab: Running Test Cases and Failure Analysis
Deliverable: Evaluation Report and Go/No-Go Decision.
Moving the solution between environments and managing it over time, including a rollback plan and user adoption.
Agentforce DX
Salesforce CLI
Git and Version Control
Monitoring and Alerts
Rollback
Kill Switch
User Adoption and Training
Outcome: A deployment and operations plan including monitoring, Rollback, and Kill Switch.
Build a complete solution simulating an organizational project – from business need to delivery – in a controlled environment with synthetic data.
Defining Use Case and Scope Selection
Agent working with Subagents and Actions
At least one Flow
Integration or code
Authorized Knowledge Sources
At least 30 Evaluation cases
Security Testing
Architecture Document and Technical Documentation
Professional Demo
Implementation Brief and Operations Plan
Outcome: Working Agent, Demo, Architecture Document, and Implementation Brief. The project is carried out in a controlled environment using synthetic data only and does not constitute a deployment for a real client, work experience, or a guarantee of placement.
Practical Outcomes
What You Will Build During the Course
Each lab concludes with a verifiable outcome, and the entire course culminates in a single brief explaining your approach to an AI solution in a Salesforce environment.
Working Agentforce Agent
An Agent with Subagents, Instructions and Actions, authorized knowledge sources, and operations based on Flow, Apex, or API.
Testing and Security Brief
Evaluation Sets with at least 30 cases in the Capstone project, Red Teaming tests, risk mapping, and Guardrails.
Implementation and Delivery Brief
Architecture document, technical documentation, professional Demo, and an operations plan including monitoring, Rollback, and user adoption.
Certificate and Certification Preparation
HPI Certificate and Professional Preparation for Agentforce Certification Knowledge Areas
Certificate of completion from HPI
Those who meet the course requirements receive a certificate of completion from HPI College. According to the syllabus, completion requirements include at least 80% attendance, a minimum score of 70 in the overall grade and final project, and zero severe security failures. Each participant is given one opportunity for corrective submission.
Professional preparation for the knowledge areas of the certification
The program prepares for the knowledge areas of Salesforce Certified Agentforce Specialist, according to the syllabus and the official name valid at the time of publication. The official exam is external and administered by Salesforce, and Salesforce certification is not automatically granted upon completion of the course.
The exam, registration for it, and the exam voucher are not automatically included in the course price, unless explicitly stated in writing for the specific cohort. The availability of Agentforce capabilities, licensing, and Entitlements varies between environments and Releases, and there is no guarantee that every tool is available free of charge. The syllabus is based on Salesforce Summer ’26 / Release 262.
Important answers regarding admission requirements, the difference from the implementer course, code, scope, tools, final project, certificate, certification, training for organizations, and suitability assessment.
The course is intended for those with relevant Salesforce background or graduates of HPI's Salesforce Implementer course. Relevant background includes familiarity with Salesforce data structure - Objects, Fields, and Relationships, basic work in Flow, User management, Permission Sets, CRUD, FLS, and Sharing, and familiarity with Sales Cloud or Service Cloud. Technical English for reading interfaces and documentation is also required. The target audience includes Salesforce Administrators, implementers and consultants, Business Analysts, information systems and CRM professionals, Flow and automation professionals, project managers, and Solution Architects. For absolute beginners, HPI's Salesforce Implementer course or a separate 40-60 academic hours Pre-Course is required first, which are not part of the 120 academic hours of the course.
The Salesforce Implementer course teaches the platform itself: CRM, data model, permissions, Flow, reports, Sales Cloud, and Service Cloud. The Agentforce course is an advanced continuation track that assumes this knowledge and focuses on AI agents: Use Case selection, Prompt Builder and Grounding, building in Agentforce Builder, integrations, Knowledge and RAG, security and governance, testing, deployment, and a final project. It does not teach Salesforce from scratch and does not replace the implementer track.
Most of the building is done with declarative tools - Agentforce Builder, Prompt Builder, Agent Script, and Flow Builder - but the track is not a no-code track. Module 07 deals with Apex, APIs, MuleSoft, and integrations, including authentication, error handling, and practice against a Mock API, and the final project requires integration or code. Previous development experience helps but is not defined in the syllabus as an admission requirement; what is required is Salesforce background and technical English.
The course includes 120 academic hours, which are 90 clock hours, and takes place over 30 sessions of four academic hours each. The material is divided into 12 modules and includes 10 labs. Hour breakdown: 75 academic hours of practice - 62.5% of the course - 27 academic hours of theory and 18 academic hours of demonstrations. To receive dates, study days, and registration details, leave your details and we will get back to you.
We build and operate automations using Agentforce Builder, Agent Script, Prompt Builder, and Flow Builder; work with information and Knowledge through Salesforce Knowledge, Agentforce Data Library, RAG, and Data 360 subject to availability; connect systems with Apex, APIs, and MuleSoft; secure using Permission Sets, CRUD, FLS, Sharing, and Guardrails; test with Evaluation Sets, Traces, and Red Teaming; and deploy with Agentforce DX, Salesforce CLI, and Git. The availability of tools and licensing varies between environments and Releases.
The final project simulates an organizational implementation and includes a working agent, at least one Flow, integration or code, authorized Knowledge sources, at least 30 Evaluation cases, security tests, an architecture document and technical documentation, a professional Demo, and an implementation file with an operating plan. The project is carried out in a controlled environment and with synthetic data only, and not on actual customer data, and does not constitute a guarantee of production deployment at a customer site.
Those who meet the requirements receive a certificate of completion from HPI College. Completion requirements according to the syllabus: at least 80% attendance, a minimum score of 70 in the overall grade and final project, and zero severe security failures; each participant is given one opportunity for corrective submission. In parallel, the program professionally prepares for the knowledge areas of Salesforce Certified Agentforce Specialist, according to the official name valid at the time of publication. The exam is external and Salesforce certification is not automatically granted upon completion of the course.
The official exam, registration for it, and the exam voucher are not automatically included in the course price, unless explicitly stated in writing for the specific cohort. In addition, the availability of Agentforce capabilities, licensing, and Entitlements varies between environments and Releases, and there is no guarantee that every tool is available free of charge. For capabilities in Beta status, an alternative practice track is defined. For precise details regarding the upcoming cohort, please leave your details.
Yes. The course is offered both as a public track for individuals and as organizational training for teams. In organizational training, the focus is on the company's Use Cases, business processes, security and governance, implementation, and value measurement; in the public track, the emphasis is on individual practice, portfolio building, and professional preparation for certification areas. If you are considering training for a team, please indicate this on the form or contact us at info@hpi.school and we will get back to you with a suitable track.
Leave your details in the form on the page. We will send you the full syllabus and have a brief conversation to review your Salesforce background and check suitability for the program. During the call, you will also receive the updated dates, study framework, and registration details for the upcoming cohort. You can also write to us at info@hpi.school.
For in-depth learning
Continue learning at the Knowledge Center
Three supplementary guides on the place of AI in the work of a Salesforce implementer, on Flow automations, and on career paths in the ecosystem.
Want to check your suitability for the course Salesforce Agentforce
Leave your details and we'll get back to you with the full syllabus, a suitability check based on your Salesforce background, and details for the next cohort. Also suitable for inquiries regarding team training within an organization.
