Building Trust in Generative AI: How Salesforce Ensures and Measures Accuracy

Generative AI is reshaping how customers find answers and solve problems - but how do you know if the response can be trusted?

At Salesforce, accuracy isn’t an afterthought — it’s foundational. Through our own journey with Agentforce on Help, we’ve developed a practical, repeatable framework to evaluate and continuously improve generative AI performance.

In this session, our Agentforce experts will share:

• How to define and measure “answer quality” in generative AI
• The framework to evaluate and improve Agentforce responses at scale
• How humans-in-the-loop (HITL) and AI work together to drive continuous learning and accuracy

You’ll leave with a clear framework to evaluate your own AI agents — and ensure every customer interaction is accurate, consistent, and built on trust.