AI in Finance: How Stripe and BNY Mellon Are Reshaping the Industry
Takeaways from Money20/20
The recent Money2020 conference provided fascinating insights into how major financial institutions are leveraging artificial intelligence to transform their operations. Two companies stood out with their ambitious AI initiatives: Stripe and BNY Mellon. Both are partnering with tech giants like NVIDIA, Microsoft, and AWS to push the boundaries of what’s possible in financial services.
Stripe: Building the Economic Operating System
Stripe is dramatically expanding beyond its payments roots with a bold new mission: becoming the “Economic Operating System” for businesses. This evolution is powered by over 100 production-grade LLMs, mostly designed for non-technical users.
A standout example is Stripe Sigma, an LLM-powered analytics tool that began as an internal solution for understanding Stripe’s own business data. Now available to customers, Sigma enables businesses to gain deeper insights into their operational data through natural language queries.
What makes Stripe’s approach particularly interesting is their focus on domain-specific LLMs built on top of foundational models. Rather than using generic AI solutions, they’re creating specialized tools that understand the nuances of business operations and financial data.
Another example is Stripe Radar, a fraud detection and prevention tool built into the Stripe payment processing platform. It uses advanced machine learning algorithms trained on global Stripe data to identify and block potentially fraudulent transactions.
BNY Mellon: The AI-Powered Custodian
BNY Mellon, the world’s largest custodian bank with over $50 trillion in assets under custody, is making equally ambitious moves in the AI space. The bank has deployed an AI supercomputer powered by NVIDIA’s DGX SuperPOD system, making it the first global bank to do so.
Their flagship AI tool, Eliza (named after Alexander Hamilton’s wife), represents a hub-and-spoke approach to AI deployment. In this model, Eliza serves as the central hub, providing core AI capabilities and governance, while different business units act as “spokes,” customizing the AI for their specific needs.
This structure, implemented at CEO Robin Vince’s direction, replaced scattered AI projects that had led to 18-month development times, bringing efficiency and central control while maintaining flexibility. The system allows employees to create “disposable” AI agents for specific tasks, from analyzing funds to customizing investment portfolios. Currently, about 14,000 employees — a quarter of BNY’s workforce — are using Eliza.
The bank’s long-term partnership with Microsoft is equally significant. Their Data Vault system, built on Microsoft Azure and launched in 2020, is a cloud-based data and analytics platform that helps investment managers and financial institutions better manage and analyze their data.
The platform enables rapid data onboarding and comes with built-in machine learning capabilities. By feeding this vast repository of financial data into LLMs, BNY Mellon is creating a powerful combination of historical financial data and modern AI capabilities.
This is particularly relevant for Corporate Treasurers who face increasingly complex challenges in managing global cash flows. In today’s environment of non-zero interest rates and market volatility, optimizing every dollar across multiple currencies and time zones has become critical. The challenge is further complicated by the financial industry’s move toward longer operating hours, driven partly by digital assets. Corporate Treasurers need sophisticated automation tools not just to manage reconciliation but to make real-time decisions about cash positioning, foreign exchange exposure, and liquidity management across global operations. Traditional manual processes and additional staffing can’t keep pace with these demands, making intelligent automation essential.
The Human Factor in the AI era
Interestingly, NVIDIA’s latest survey revealed a significant shift in industry priorities. For the first time, attracting and retaining talent has surpassed data as the primary challenge for organizations. This highlights an important reality: while AI tools are becoming more sophisticated, human expertise remains crucial.
Much like a modern molecular gastronomy kitchen needs more than just line chefs, successfully implementing AI in banking requires a diverse “kitchen staff.” Data scientists alone — acting as sous chefs or line cooks — won’t be sufficient. Organizations need a complete team: executive chefs focused on change management, specialist chefs dedicated to risk and cybersecurity, and even a dedicated “AI culinary school” for talent training. This new kitchen setup will likely see a blend of human expertise and AI “robot chefs” handling different aspects of operations, with humans increasingly focusing on quality control and strategic oversight rather than routine tasks. The key is building a balanced team that can navigate both the technical and organizational challenges of AI transformation.
The Journey Ahead of Us
Both Stripe and BNY Mellon’s approaches suggest we’re entering a new phase in financial AI. Much like the evolution of baking from the invention of baking powder to molecular gastronomy, [1] the banking industry’s AI journey is progressing from basic efficiency gains to potentially revolutionary new experiences.
While the current scaled use cases are primarily focused on information retrieval and code writing — similar to how baking powder first brought efficiency and consistency to baking — we’re seeing the early signs of a more transformative future. Just as molecular gastronomy combines art, science, and experimentation to create entirely new culinary experiences, financial institutions are beginning to:
- Develop domain-specific AI models trained on proprietary data (their core ingredients)
- Create tools that empower non-technical users (their new equipment and techniques)
- Build integrations with existing systems and workflows (their kitchen lab infrastructure)
The journey of GenAI in Banking is progressing from what baking powder did for baking in the 19th century (mostly efficiencies) toward something more revolutionary — the equivalent of molecular gastronomy in finance. This isn’t just about precision; it’s about transformation, experimentation, and combining art and science to create entirely new financial experiences.
For financial institutions watching these developments, the message is clear: AI is no longer just about efficiency gains — it’s about fundamentally reimagining how financial services can be delivered in the digital age. Just as molecular gastronomy transformed desserts, AI has the potential to transform banking from transaction processing to something we haven’t yet imagined. The question is no longer whether to adopt AI, but how to begin this transformative journey from traditional banking to its “molecular gastronomy” equivalent.
📌 Subscribe to my YouTube Channel with my insights and industry leader interviews. New video every Wednesday: https://www.youtube.com/EfiPylarinou
📌 Twitter: https://twitter.com/efipm
📌 Apple Podcast 👉 https://buff.ly/3P1Cp1Z
📌 Spotify Podcast 👉 https://buff.ly/3xPyWaV
📌 Linkedin: https://www.linkedin.com/in/efipylarinou/
📌 TikTok: https://www.tiktok.com/@efiglobal
