Platform, Automation, and AI (Agentforce) for Small Business
Declarative automation first
Salesforce automation is layered. Know the tools and when each applies: Flow Builder is the primary automation tool (screen flows, record-triggered flows, scheduled flows); approval processes route records for sign-off; validation rules block bad saves. Older tools like workflow rules and process builder are legacy - questions often test that Flow is the recommended replacement. A frequent trap: order of execution, where validation rules and before-save flows run before the record commits, and after-save flows run afterward and can update the record again.
The platform building blocks
Understand that custom objects, fields, page layouts, and Lightning apps are declarative configuration, while Apex and Lightning Web Components are code. Know that sandboxes are copies of production for testing and that changes move through change sets or deployment tools.
Agentforce and AI
Agentforce lets you build AI agents grounded in your Salesforce data. Key concepts: an agent has topics and actions; topics scope what the agent handles; actions invoke Flows or other logic; grounding ties responses to your records so answers are accurate. Einstein features like predictive scoring and generative summaries are configured, not coded. Traps: agents need permission-aware access to data, and ungrounded agents hallucinate - grounding and guardrails matter.
Worked rule of thumb
For any 'how should we automate this?' question, prefer declarative Flow over code, and before-save flow for same-record updates, after-save for related records. For AI, always ask what data grounds the agent and whether the running user can access it.