AI on Salesforce and the Einstein Trust Layer

Agentforce & AIAI on Salesforce and the Einstein Trust LayerHow 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.Beginner · about 60 minutesWhat you will learn

Trace the evolution from Einstein predictive AI to Einstein GPT, Einstein Copilot and Agentforce, and use current product names

Distinguish predictive, generative and agentic AI and pick the right one for a business problem

Explain tokens, context windows, temperature, hallucination, grounding/RAG and prompt injection in developer terms

Describe each stage of the Einstein Trust Layer request and response path

Call Salesforce-managed or BYO models from Apex (aiplatform.ModelsAPI) and REST, including chat and embeddings

Enable Einstein generative AI and Agentforce in an org and understand licensing and data residency at a high level

Contents

From Einstein to Agentforce: a short history

Predictive vs generative vs agentic AI

LLM fundamentals for developers

The Einstein Trust Layer

Models on Salesforce

Licensing and credits (high level)

Enabling Einstein generative AI and Agentforce

Data residency and the geography of AI

Worked example: an AI-generated account brief with safety checks

Best practices

Common pitfalls

Hands-on exercises

Key takeaways