Salesforce Agentforce: Learn to Build AI Agents (Fast)

Set up an Einstein AI enabled developer org with Data Cloud and Agent Force, activate features, create an agent user, enable the default agent, and test account data actions.

Create the Agent Object Access permission set for Einstein Agent and grant read, create, and view all fields on accounts, cases, contacts, opportunities, tasks, and conversation entries, avoiding delete rights.

Enable the digital experience option in Salesforce, create and publish an Experience Cloud site for agent-assisted customer support, and prepare to embed the agent force agent in future sessions.

Create a basic service agent with Agent Builder to let customers view open cases, configure case management topics, and set the role for customer support.

Salesforce Agentforce enables building actions and flows to retrieve open cases by customer email, and present case number, subject, description, and status.

Test the agent’s ability to identify prompt intent and execute actions, such as retrieving open cases by email. Verify itemized fields like case number, subject, created date, and status.

Enable omnichannel on the experience cloud site, set presence statuses, route to the service agent via a dedicated service channel, and publish the embedded service deployment for testing open cases.

Build a customer support agent on your experience cloud site with Salesforce Agent Force, capable of fetching case details, adding comments, updating fields, recommending knowledge articles, and creating cases.

Set up the pre-chat email collection and omnichannel flow in Salesforce to identify customers, link to the contact via the messaging session, and route to the Agent Force assistant.

Configure a service agent to fetch case records by case number, presenting subject, description, contact email, created date, last modified date, with guardrails against hallucination and testing in omnichannel.

Learn to add a new comment to an existing case by creating a dedicated topic, requiring a case number, verifying the case, gathering the comment, and confirming before creation.

Learn to update existing case records via an AI agent flow using a case number, selectively updating fields, preserving existing values, and presenting updated case summaries in experience cloud.

Create a new case from a customer's query on the experience cloud site, using the messaging transcript to generate a concise subject and detailed description after confirming details, without diagnosing.

Set up knowledge for Salesforce Agentforce by creating and publishing knowledge articles, organizing data categories, granting Einstein agent read access, and preparing a knowledge data library for the agent force.

Explore the data library in agent force, a knowledge hub that connects Salesforce knowledge articles, cases, and custom data to power AI agent responses during chats.

Explore how to enable a Salesforce agent to answer customer questions by searching lightning knowledge articles and returning the article content, using Agent Force to fetch, summarize, and reference knowledge.

Test your customer support agent with batch tests, configure simulated environments using initial context variables, and generate csv-backed test cases to evaluate completeness, coherence, conciseness, and latency.

Explore transferring a conversation from an AI agent to a live support rep via omnichannel, configuring inbound and outbound flows to route to a queue while preserving context.

Extend Salesforce's default agent to build a sales assistant that schedules today's meetings, highlights opportunities closing this week, scores opportunities, suggests next actions, and qualifies leads with real-time CRM updates.

Discover how to build a Salesforce agent that fetches today’s tasks and events from standard CRM objects, using an invocable Apex action to display a structured, human-readable schedule.

Expand your use case to include open opportunities closing this week, highlight overdue tasks, and present opportunities in a separate, urgent section with clear actions and record context.

Build an AI agent to identify leads not contacted in the last seven days, include lead and opportunity data, and output a cohesive, clearly formatted list with urgency cues.

Show the opportunity, its score, and a recommendation by the current opportunity ID, with automatic retrieval, and compute the score from amount, stage weight, and days open to guide closing.

Build a lead qualification assistant that scores leads via four questions (0/10/25), logs data into Salesforce, and updates the lead score field or uses a custom score.

Treat your agent like a new hire by feeding high quality data and clear voice guidelines, use guardrails and handover for unknowns, and start small to iterate toward context-aware responses.

Earn the agent for specialist certification by mastering agent force concepts, prompt engineering, service cloud, sales cloud, and data cloud, with hands-on practice, trailhead prep, and community resources.

Compare Einstein Copilot and Agent Force: use Copilot for user-guided CRM tasks and drafting actions, while Agent Force autonomously executes multi-step automations across apps and data.

Explore the future roadmap of Agent Force, featuring custom templates, versioning, and safe cloning like GitHub for AI agents, plus next-gen chat, voice, live translation, and multi-agent orchestration.