The Salesforce Admins Podcast Why Secure AI Starts With You: What Admins Must Know About Agentforce
5 snips
May 22, 2025 Sri Srinivasan, Senior Director of Information Security at Salesforce, dives deep into the world of AI security. He emphasizes the pivotal role of admins in securing AI experiences through Agentforce, likening it to driving a sports car without brakes. Sri shares five essential questions admins should consider when implementing AI, urging proactive security measures from the start. He also highlights the importance of ethical considerations and community engagement, reminding listeners that effective user interactions and strategic planning are key to navigating AI's complexities.
AI Snips
Chapters
Transcript
Episode notes
Shared Responsibility in AI Security
- AI security is a shared responsibility between Salesforce and admins.
- Security should be integrated into agent roles, data access, actions, guardrails, and channels from the start.
Set Agent Guardrails Effectively
- Set clear agent guardrails using natural language instructions.
- Prevent agents from asking for unnecessary sensitive info or acting without user verification.
Admin 'Aha' With Agent Reasoning
- Admins have aha moments when they see the agent builder's reasoning explaining actions.
- This understanding empowers them to tweak prompts and crack the code of agents.