As AI moves toward agentic systems capable of performing tasks with minimal human intervention, guardrails setting boundaries and human verification of potentially unsafe output become crucial for enterprise applications, according to Salesforce executive vice president and chief scientist Silvio Savarese. In this episode of Tech Disruptors, Savarese and Bloomberg Intelligence senior technology analyst Anurag Rana take an in-depth look at the differences between gen-AI assistants and the agents, which the company announced at its Dreamforce 2024 conference. Other topics include the application of small vs. large language models, enterprise AI vs. that for the general public and security.