Agentforce & AI

Agentforce & AIBuild autonomous AI agents: Einstein Trust Layer, topics & actions, Prompt Builder, Agent Script, Agentforce DX, testing, Agent API, MCP and multi-agent systems.

AI on Salesforce and the Einstein Trust Layer: How Salesforce went from predictive Einstein to autonomous Agentforce agents, the LLM fundamentals every developer needs, how the Einstein Trust Layer protects every generative request, and how to call models directly with the Models API from Apex and REST.

Agentforce Concepts and Architecture: What an Agentforce agent really is: the reasoning engine, subagents (formerly topics), actions, variables, knowledge, channels, running users and guardrails, plus the metadata that represents an agent in source control and the design principles that make agents reliable.

Prompt Builder and Prompt Templates: Design, ground, test, version and invoke prompt templates: template types, merge fields, Flow and Apex grounding, Data 360 retrieval, model selection, and calling templates from Lightning, Flow, Apex, REST and agents, with prompt-engineering techniques that produce reliable output.

Building Agent Actions: Design and build the tools your agents call: Apex invocable actions, autolaunched Flows, prompt template actions and API actions, with LLM-friendly contracts, confirmations, structured outputs, Custom Lightning Types, security, idempotency and testing, plus a complete Order Status and Return subagent.

Agent Script and the New Agentforce Builder: Make agents predictable with hybrid reasoning: learn the Agent Script language (blocks, variables, subagents, logic and prompt instructions, actions, conditions, transitions and lifecycle hooks), build a complete order-support agent with deterministic identity verification, and manage .agent files with Agentforce DX.

Agentforce DX: Pro-Code Agent Development: Build, preview, publish, test and ship Agentforce agents like any other code: the sf agent CLI commands, agent specs, authoring bundles with Agent Script, agent metadata types, retrieve/deploy, packaging, CI, the Agentforce DX VS Code extension and Agentforce Vibes. Ends with a full walkthrough from an empty project to a deployed, tested agent.

Testing, Evaluation and Observability for Agents: How to prove an agent works and keep it working: why agent testing is evaluation rather than assertion, unit-testing actions, preview traces, Testing Center suites (AiEvaluationDefinition) and the newer Agentforce Studio runner, sf agent test in CI, red-teaming and guardrails, and production observability with Session Tracing, audit data, health monitoring, cost and quality KPIs.

Agent API and Headless Agents: Run Agentforce agents without Salesforce UI: the Agent API (sessions, synchronous and streaming messages, message types, context variables) with complete TypeScript and Python clients, invoking agents from Apex and Flow, the Agentforce Mobile SDK, Messaging for In-App and Web, Slack and Voice channels, plus security, error handling and limits.

MCP, A2A and Multi-Agent Systems: Connect Agentforce to the wider agent ecosystem: Model Context Protocol fundamentals, Salesforce as an MCP client and as a host of MCP servers, the Salesforce DX MCP server and its retirement, building your own MCP server over Salesforce APIs in TypeScript, the Agent2Agent protocol, multi-agent orchestration patterns, MuleSoft Agent Fabric, Heroku AppLink and the security model that keeps it all safe.

With Sharing needs JavaScript to run. Turn it on to browse the catalogue and read lessons.