This episode explores the current state and future potential of AI agents, specifically focusing on their application in customer experience. Against the backdrop of differing academic and industry definitions of "agents," the guest, Bret Taylor, categorizes them into personal, persona-based, and company agents. More significantly, he highlights the practical viability of company agents, exemplified by Sierra's work with Sonos and SiriusXM, which leverage AI to handle customer interactions, from onboarding to troubleshooting. For instance, Sierra's platform allows companies to define goals and guardrails for their AI agents, balancing automation with brand consistency and compliance. The discussion then pivots to the challenges of integrating foundation models into company agents, emphasizing the need for systems capable of orchestrating complex processes and handling actions beyond simple question-answering. In contrast to the prevalent "retrieval augmented generation" approach, Taylor advocates for a platform that enables the orchestration of complex processes, incorporating both goals and guardrails. This approach aims to harness the creativity of AI while maintaining control and ensuring brand alignment. Emerging industry patterns reflected in this discussion suggest a move towards outcome-based pricing models and a shift in the relationship between software vendors and their clients, focusing on measurable value delivery.
Sign in to continue reading, translating and more.
Continue