Mastering Salesforce AI: Agentforce and Prompt Templates

Learn the fundamentals of agent for studio, distinguish service and employee agents, and understand running user, Einstein agent license, and key builder blocks such as agent detail and welcome message.

Explore agent script components, including agent detail (name, api name, running user, description), agent level and sub-agent instructions, language settings, and system messages.

Create and configure a service agent in agent script, assign an Einstein user, test in live and simulate modes, and manage sub agents to support soft drink orders.

Configure the start underscore agent to route to the correct soft drink sub-agent, then implement an auto launch flow action with rating input to fetch records.

Finalize the sub-agent and custom action, test agent populated inputs via llm slotting, and configure deterministic flows with variables, permissions, and versioning for soft drink records by rating.

Make the soft drink agent deterministic by introducing a rating variable and a set rating action, using if conditions to ensure rating is available before querying records.

Learn verification for the verify customer sub-agent by sending an OTP from the authentication key to the user, then compare user key with authentication key and route via agent router.

Explore how agent script enables deterministic, hybrid reasoning with context engineering, conditional logic, and action filters to safely route between subagents and control prompts.

Learn essential agent script terminology and syntax, including the at rate symbol for actions and variables, template instructions, pipe for multiline text, and inline run action with outputs.

Deploy a service agent to a customer community site with Agentforce, comparing direct routing and omni channel flow, and configure messaging channel, routing, and fallback queue for reliable external deployments.

Deploy a Rumi ai agent to a customer site using omni-channel flow, routing configurations, fallback queues, and an embedded messaging channel with a seamless community site integration.

Route work from ai agents to human agents via omni channel in service cloud, using messaging service channels and present statuses to optimize simple queries and a refund.

Design and deploy an omni channel flow to route from the AI agent to human agents, pass the messaging session ID, create and link cases in queue, and track KPI.

Understand why Salesforce needs department-specific employee agents for internal users. Assign department-specific agents to sales, finance, or library, not as a single service agent.

Assign sales and lead employee agents to user groups via permission sets, enabling agent access for each user, and test with an Apex class creating soft drink orders.

Enable Slack integration with Agentforce coworker to read recent Slack messages, channels, and user details from Salesforce data cloud, while noting read-only access and no posting capabilities.

Explore how Salesforce agents trigger from conversations or data changes, plan via the Atlas reasoning engine, and execute actions across planning and execution phases, transferring tasks to humans when needed.

Explore Einstein AI architecture, from sales, service, marketing, and commerce agents to Agent Force, guardrails, and the Einstein Trust layer with Atlas Reasoning Engine guiding autonomous, data-triggered, and assistive agents.

Explore how the Einstein Trust Layer uses data masking to protect sensitive data and enforces zero data retention when using generative AI with external LM providers.