💬 In plain words: A user needs two separate grants before an agent is useful to them. One lets them use the agent at all. The other lets the agent reach data. They fail in completely different ways. 📌 Example: A Meridian rep is granted every object the agent touches, but not Agent Access. The agent simply does not appear for them. Nothing is wrong with the data model. 🎬 Real-Life Example: The Agent Nobody Could Find The Old/Bad Way: An admin built an agent, activated it, tested it in preview, and told the team it was live. Nobody could see it — including the admin, outside preview. Why this fails: Agent Access is a separate grant inside a permission set. Building an agent does not grant you the right to use it. Preview bypasses this, which is exactly why the gap goes unnoticed. The New/Good Way: 1. In the permission set, find the Agent Access section. 2. Add the specific agent to the enabled list. 3. As...
💬 In plain words: About twenty words carry most Agentforce conversations. Learn these once and every module reads faster. Each entry points at the module that covers it properly.
Concept
• Agent: software that reads language, decides, and acts.
• Employee Agent: serves staff, runs as the logged-in user.
• Service Agent: serves customers, has its own user.
• Agent Router: picks one subagent per message. Was the Topic Selector.
• Subagent: a specialist inside an agent. Was called a topic.
• Action: something the agent can actually do. Five kinds.
• Tool: an action the model is allowed to choose for itself.
• Agent Script: the language agents are written in.
• Grounding: feeding real Salesforce facts into the prompt.
• Prompt template: a saved, reusable, grounded prompt.
• Data Library: the store of documents an agent can search.
• Retriever: the thing that searches it and returns chunks.
• Chunk: one small piece of a document.
• Trace: the record of what actually happened in one turn.
• Einstein Trust Layer: the platform-level protections underneath.
• Digital Wallet: where you see credit consumption.
🧠 Say the new words: subagent not topic, Agent Router not Topic Selector, Data 360 not Data Cloud. Old words date you instantly.
🧭 360 Card — Key Terms in Plain Words
Rule: learn the vocabulary before the features. Wrong words undo correct answers.
Gain: you sound current, and you can read Salesforce documentation without translating.
Price: the vocabulary moves. Some of these will change again.
Limits: the interface still says "Topic" in places. Same thing.
Mirror — learning features first: you can describe what a thing does and still name it wrong.
Core
Q&A
Q:
What is the difference between an action and a tool?
🎯 Say this first: An action is a capability. A tool is an action the model is allowed to
choose for itself.
A: They are the same object seen from two
sides.
• Anything in the plain actions list runs
because your logic called it. That is deterministic.
• Anything exposed in reasoning actions is a
tool. The model reads its description and decides whether to use it.
• So the same underlying Apex or Flow can be
either, depending on which list you put it in.
• That choice is a design decision, not a
technicality. It decides who is in control of that step.
📝 2-Minute Self-Check
Q1. A customer types five different phrasings of "my power is out". Which layer handles that?
A1. Routing. The Agent Router classifies all five to one subagent.
Q2. Does Agentforce require Data 360?
A2. Only for grounding on unstructured content. Record, Flow and Apex grounding need none.
Q3. What was a subagent called before April 2026?
A3. A topic. The Topic Selector is now the Agent Router.