As someone completely new to the Salesforce world, I know exactly how it feels to be faced with a powerful new technology like Agentforce. When I was asked to explore what it means to build an artificial intelligence (AI) agent, my first reaction was to look over my shoulder to see who they were actually talking to.
For me, the idea of creating something with AI felt like being asked to cook a five-course meal when I’d only ever made toast. It seemed like a job for seasoned experts, not beginners.
But here’s the thing: I’m not the only one who’s felt that way. We’ve seen Trailblazers from all different backgrounds show up to Agentforce NOW workshops with the same uncertainty. What I’ve learned from your questions and exploring the platform myself is this: You don’t need a decade of experience or a technical title to get started.
So, if you’ve been curious about Agentforce but are unsure where to begin, let’s go through the learning journey together. I’ve organized the top questions we hear in Agentforce NOW workshops into three phases to help you get started. Then, when you’re ready for your own ‘aha!’ moment, I highly recommend attending a workshop to get hands-on and see it all live.
The journey for so many learners starts with one of the most powerful and common concerns, which I definitely shared: Is this even for me? It’s a question that sits at the heart of all the practical ones that came up first.
The good news is, you can start building and exploring today. Salesforce provides free environments made specifically for this purpose. Besides joining an Agentforce NOW workshop, your best options are:
That’s a great question. Remember that your AI is only as good as your data. Here’s how to think about their relationship.
Think of Agentforce as the smart “brain” that can perform tasks and take action. Think of Data Cloud as the rich, unified “memory” that gives that brain a deep and contextual understanding of your customer.
Data Cloud is what unlocks an agent’s full potential. For instance, an agent might be able to summarize a single case on its own. With Data Cloud, it can access and harmonize data from multiple systems to give a complete picture of related data for the task, allowing it to provide much smarter and more relevant assistance. Remember that agents are more successful when given more context about the problems they’re trying to solve, and Data Cloud provides the best solution for this.
The good news is that Salesforce Foundations, the free add-on available to many customers, is designed to include both Agentforce and auto-configured Data Cloud. They’re bundled together because they’re truly better together.
The simple answer is no, it can’t. An agent’s security is based on your own. It can never see any data that the person using it doesn’t already have permission to see. Trust is our #1 value at Salesforce. The permissions and user access you put in place at your organization all carry over to the way agents access data and act.
With Agentforce, you’re always in control. For any given task, you decide exactly what information to show the agent. It’s not free to browse your systems—it only sees the specific data you allow it to see for that job.
Once I was in, my next step was to understand the ‘why’ behind the ‘what.’ Before you can build anything great, you have to understand the foundational design. The following questions cover that core logic that makes everything else work.
Think of it like a skills menu for your agent. A topic is a category on that menu, like “Sales” or “Customer Service”. An action is a specific skill listed under that category.
For example, under your Sales topic, you’d have specific actions like “Summarize an Opportunity” or “Update a Lead’s Status”.
So, how does the agent pick the right action from the menu? That’s where AI comes in. It uses natural language processing to understand the intent behind your request. It reads what you wrote, figures out what you’re trying to accomplish, and then finds the specific action that’s the best match to get the job done.
That’s a perfect question because it gets to the core of how to design a smart agent. The most important thing to know is that topics aren’t about chunking data or articles.
Another way to think about it is like chapters in a cookbook. A cookbook isn’t just a pile of ingredients (the data)—it’s organized by what you want to make.
Chapters, like “Appetizers” or “Main Courses”, are your topics. They’re logical categories for a type of work.
The specific recipes inside each chapter, like “Make Bruschetta” or “Grill a Steak”, are your actions. They’re the individual tasks the agent knows how to complete.
So, you should always define your topics based on the function you want the agent to perform, not the data you want it to reference.
A great way to think about this is like hiring employees for different departments in your company. You hire people for specific roles, and you should think about your agents the same way.
You’ll often hear about two main types of agents based on their ‘job function.’
This brings us to when you need more than one. You’d create a separate Service agent for your customer support team and a separate Employee agent for your internal sales team, for example.
The most important reason for this is to control data access and job function. Your customer-facing Service agent should never have access to sensitive internal employee data, and your internal Employee agent doesn’t need to see private customer service cases.
So, best practice is to create different agents for different, distinct job functions—especially when separating internal (employee) and external (customer) tasks. Learn more in the Employee Agents: Your User Support Advantage blog post.
This next part is the ‘aha!’ moment that changed everything for me. It’s when I realized this phase isn’t just about using existing tools—it’s about the surprising power you have to create new ones.
My biggest lesson in all of this is that ‘I don’t know’ is a great place to start, because it’s always followed by ‘let’s find out.’
Yes, and this is where the platform truly opens up. You aren’t limited to standard actions.
You can create your own custom actions using the Salesforce tools you already know. For most needs, you can use Flow to build powerful automations with clicks, not code. If you have developer resources, you can also use Apex. This allows you to perfectly tailor the agent to your company’s unique processes.
Currently, an agent is designed to interactively work alongside a person, rather than run completely on its own in the background.
This means that right now, an agent is always started by a user. You can’t call it automatically from a flow or Apex trigger to process records.
However, the team at Salesforce knows how powerful this would be, and it’s something they’re actively exploring for future releases—so stay tuned!
That’s a great question, and the answer is one of the most exciting things about the platform: Agents are designed to meet your users and customers exactly where they already work.
This flexibility means you can deploy different agents in a variety of places. For example, you can:
The goal is to bring helpful, trusted AI directly into everyone’s flow of work, making it a seamless and natural part of their day.
Once you’re in a workshop and building, you might start thinking about what’s next. Here are a couple of common questions we often see from learners who are ready to take their skills to the next level.
While the workshop isn’t an official certification prep course, it provides a strong, hands-on foundation with the core concepts that are essential for the exam. It’s a fantastic first step in your certification journey.
For more details on diagnosing common issues, check out our 6 Tips To Help You Troubleshoot Agentforce With Confidence.
My biggest takeaway from this whole experience is simple: The best way to get rid of self-doubt is with action. You can read articles all day, but the real learning starts the moment you get your hands dirty.
So, let’s get you started.
The best thing you can do is sign up for a free, interactive Agentforce NOW workshop. See for yourself how achievable it is to implement agents at your organization. Your ‘aha!’ moment is waiting.
I can’t wait to see what you build!
Artificial intelligence (AI) agents are revolutionizing app development. Join us for the Agentforce Virtual Hackathon to push the boundaries of AI-powered agent technology and build groundbreaking solutions on Agentforce for the chance to win a $50,000 Grand Prize. With Agentforce, Salesforce is leading the way in the biggest technological revolution in decades. Agentforce is the […]
Editor’s note: This post was updated on March 25, 2025, with the latest information and resources. As Salesforce Admins, we have clear core responsibilities that define our day-to-day roles. We manage users, data, and product, provide actionable analytics, and oversee security. But we’re currently in the third wave of the artificial intelligence (AI) revolution: contextual […]
This article was co-authored by Salesforce Program Manager Anna Bromley and Salesforce Solution Architect Jeff Kallenbach. From Hackathon pitch to working demo in 14 days Let me take you back to June 2025. The lights were bright overhead, and the low hum of TDX London buzzed around. Jeff and I were on stage, about a […]
