In this episode of Next Frontiers of AI, host Scott Hebner is joined by Peyman Parsi, Senior Principal for Financial Services at MongoDB, to examine a critical industry challenge: why roughly two‑thirds of AI projects in financial services stall before reaching production, and why those that do often fail to scale with the business. With the advent of agentic AI and its higher‑stakes use cases, this challenge is only becoming more pressing.
The conversation examines the organizational, technical, and trust barriers that prevent promising proofs-of-concept from scaling, ranging from legacy infrastructure and governance gaps to rising costs, bias, and unclear ROI. Scott and Peyman discuss how financial institutions can overcome these obstacles by adopting trusted, agentic architectures built on strong data foundations. These architectures enable features like vector search, persistent memory for AI agents, domain-specific embeddings, and seamless system integration, transforming AI from an isolated initiative into an enterprise-wide capability.
The discussion also looks at the evolving GenAI landscape in financial services, from corporate GPTs and vertical use cases to the trade-offs between high-risk automation and human-in-the-loop models. With real-world examples in fraud detection, portfolio management, customer personalization, and operational AI assistants, this episode offers actionable insights for leaders aiming to turn AI hype into measurable ROI in high-risk, high-reward areas like finance.
