Baking GenAI in Banking
In the evolving landscape of financial services, we stand at the cusp of a transformative phase with Generative AI. Just as baking powder revolutionized baking in the 19th century, GenAI promises to transform Banking from Digital to Intelligent. Drawing insights from the Money20/20 panel discussion I led with leaders from Microsoft, Citizens, and iGenius, this article presents a framework comparing the journey of Baking GenAI in Banking to the evolution of Baking.
From selecting core ingredients to building a specialized kitchen lab, we explore how banks can progress from basic efficiency gains to creating revolutionary new experiences — their own version of molecular gastronomy in financial services.
The Evolution of Banking Technology: From Credit Cards to GenAI
It is only seven decades since the first credit card was launched in the 1950s and five decades since electronic trading became a thing. Mobile was baked into banking at the beginning of this century, and now we stand at the early stages of the AI technological adoption phase. This transformation promises to elevate Financial Services from merely Digital to truly Intelligent.
At Money20/20, the evidence of this transformation was clear. Panel discussions featured actual GenAI use cases from both incumbents (e.g., BNY Mellon) and grown-up Fintechs (e.g., Stripe), alongside several startups powered by NVIDIA’s AI platform and their partner ecosystem of cloud providers.
The Baking Powder Moment: GenAI as a Fundamental Innovation
One of the milestone innovations in baking was the invention of baking powder in the mid-19th century. When baking powder was invented, it revolutionized the industry by making baking more efficient, easier, and consistent for cakes to rise — benefits that applied equally to both retail and enterprise settings. This analogy perfectly illustrates where we are with GenAI in banking today.
As highlighted during our Money20/20 panel discussion on “The Killer GenAI Cake: mixing Big Tech, FinTechs, Unicorns, and Banks,” we find ourselves at a similar inflection point. Just as baking powder standardized and improved efficiency in baking, GenAI is poised to do the same for banking operations.
However, the vision extends far beyond mere efficiency gains. Subsequently, baking evolved with new tools — mixers and ovens — that added additional efficiencies and automation. We now have cake mixes that can be used instead of baking from scratch, allowing for customization and catering to special requirements like gluten-free or dairy-free cakes. Similarly, in banking, we’re seeing the early stages of customization and specialization in GenAI applications that don’t require baking from scratch.
The key difference is that while GenAI might appear to be just our “baking powder moment,” it actually holds the potential for much more transformative change. As Tyler Pichach from Microsoft emphasized during our panel, “If you can get your data state set up in the right way, this is where the magic can happen then to unlock all of the great use cases, but you’re not going to get to that unless you have a core governance structure.”
Preparing Your Core Ingredients: Data Strategy
To start on the journey of Baking GenAI in Banking, bankers must first choose their core ingredients — their equivalent of flour, sugar, eggs, and butter. Just as a baker selects ingredients based on their final product, the mix of core ingredients will depend on the strategic positioning of the business.
Core Components
For a universal retail consumer banking business, these ingredients might include:
- Transaction records (your flour)
- Geolocation data (your sugar)
- Product descriptions (your eggs)
- Customer feedback (your butter)
For an investment-focused business, the recipe changes to:
- Market prices
- Financial reports
- News articles
- Voice recording metadata
As Uljan Sharka from iGenius emphasized, “Success is in the details and like baking, it’s all about the process. It’s about understanding how to blend ingredients together. It’s also about understanding which flour to use. Can we use a flour that’s good for all cakes? Do we need something that is more domain specific and fine-tuned to the area of financial services?”
The quality and accessibility of these ingredients are paramount. As we highlighted during our panel, even if you have quality core ingredients, if your flour and sugar are mixed up, your butter melted, and your eggs broken — or in banking terms, if you can’t access your data in the required conditions at the right time — you can’t even begin to bake. Imagine if you are handed the best ingredients but they are lumped in a bowl, and you can’t separate them.
Staffing Your Kitchen Lab: The Human Element
The next crucial step in Baking GenAI in Banking is staffing your ‘kitchen lab’.
Kitchen Staff Organization
Even though right now you may think it suffices to get the right mixers and ovens from cloud providers like Microsoft, and appoint a few `sous chefs` to manage the baking processes (i.e. implement the strategy) and a few `line chefs`that execute recipes and handle operational procedures (i.e. some data scientists); that is only part of what is needed.
As Christine Cavallo and Tyler Pichach highlighted during our panel Baking GenAI in Banking is less of a technology problem and more of a people process change management challenge. They envision, the kitchen lab for Baking GenAI in Banking, eventually including humans, chefs, in addition to agent robot chefs. These robot chefs may be taking care of things that need absolute hygiene and real-time 24/7 precision and the human kitchen staff maybe doing more quality control than the actual cooking.
For now, as Christine Cavallo emphasized Bankers need a chef that’s focused on change management. Another chef focused on risk and cyber. A dedicated AI culinary school for talent training.
Bankers need to realize that Data Scientists alone (in the roles of sous or line chefs) won’t be able to successfully Bake GenAI in Banking.
The vision for the future kitchen lab includes both human chefs and agent robot chefs, with automated systems handling tasks requiring absolute precision while humans focus on quality control and innovation.
Tools and Techniques: Technology Infrastructure
Just as modern baking relies on professional-grade mixers, ovens, and precise measurement tools, Baking GenAI in Banking requires a robust technology infrastructure. However, as our panelists emphasized, getting the right equipment is only part of the solution.
Cloud providers like Microsoft are offering increasingly sophisticated “kitchen equipment” for banks. Tyler shared Microsoft’s own experience: “We get to test things early and figure out what works and what doesn’t work. And then we get to take those learnings or those recipes and be able to talk to our partners and our customers about how they can advance what they do leveraging it.”
A successful example of this kind comes from Microsoft’s contact center implementation. Microsoft has implemented genAI to empower its 45,000 contact center agents worldwide. Initially estimated as a two-year project as it included dealing with multiple languages and specializations. The implementation with GenAI took only four months as the user adoption was high due to automated note-taking and the appropriate setup (staff, tools, recipes) to leverage the natural language knowledge search capabilities of LLMs.
From Basic Baking to Molecular Gastronomy: The Future Vision
While we might be in the “baking powder era” of GenAI in banking, the vision is clearly set on achieving molecular gastronomy-level innovation — creating combinations and experiences that were previously impossible.
Uljan Sharka, founder of igenuis has developed a Betty Crocker kind of cake mix for asset & wealth managers. Their cake mix has precision baked-in.
Uljan Sharka says: “We focus on domain-specific AI. So to put it in terms of cakes, we provide a very special Italian tiramisu. Which is basically serving asset managers and wealth management.”
This specialized approach demonstrates the future of GenAI in banking. Uljan believes we will be seeing more Domain-specific solutions (think of them as Betty Crocker cake mixes), with zero tolerance for errors (high-quality control), enabling enhanced user experiences (potential for delightful decorations and presentations).
Such a domain-specific LLM allows asset/wealth managers to create their own layered cakes but not from scratch.
Key Takeaways
The journey from basic GenAI implementation to true innovation in banking requires careful planning, the right ingredients, skilled staff, and appropriate tools. As we progress from our current “baking powder moment” toward the equivalent of molecular gastronomy in banking, organizations must maintain a balance between innovation and reliability, automation and human oversight, and efficiency and customization.
Use the Kitchen Readiness Assessment Framework below for your journey to Bake GenAI in Banking.
I am an Microsoft ambassador. Opinions in this article are my own.
