As AI agents multiply, so do the risks. Discover how Salesforce plans to treat every agent, first or third-party, as a first-class citizen with granular authorization and traceability.
Product: Salesforce Platform
Session Level: Intermediate
Session Summary: The session introduced Salesforce's innovative Agent Identity feature, addressing critical security challenges posed by AI agents operating within enterprise environments. Agent Identity establishes each agent as a verified user with unique credentials, enabling full traceability, least privilege access, and independent control to mitigate risks such as over-permissioned agents, compliance issues, and lack of accountability. The feature incorporates dual attribution, linking agent actions to both the agent and a human accountable user, ensuring transparency and accountability in all operations. Currently in pilot, with general availability planned for November, this advancement exemplifies Salesforce's commitment to enhancing security and governance as AI agents become integral to enterprise workflows.