AI Will Transform Finance, Fintech Will Transform Itself

April’s tariff fiasco has laid bare the fact that, while AI is a promising technology with the potential to significantly impact operating costs, much of the global economy is rooted in physical inputs and goods, sourced and sold around the world. In 2024, AI captured investor imaginations. In 2025, anticipate a more surgical integration of AI with real-world systems, sparking innovation built on the increasingly sophisticated replication of agentic workloads, or else circumventing the need for AI altogether. Fintech’s near-future is coming into focus, and it looks evolutionary.

Celebration of LLMs has been replaced this year by GenAI’s promise to power AI agents, little kernels of intelligence designed to reason, make decisions and execute tasks autonomously. Agentic AI is especially compelling to financial institutions because much of the financial sector is about following loose processes linked together by judgment-based tasks. I’ve written previously about data engineering and agentic operating systems for all types of businesses, and how the implementation of contemporary data engineering helps to elucidate a specific business ontology, against which the promise of independent agents, capable of proactive interaction, can be unlocked.

This presupposes a whole number of other technologies, some of which were fairly esoteric prior to the AI boom; for example, digital twins and synthetic data. Long used in manufacturing and urban planning, the concept of a digital twin can now be applied to retail and financial institutions, where every aspect of the entire business can be replicated through individual AI agents as a digital entity. This involves combining business data and LLMs to create agents that represent different aspects of the business, such as stores, distribution centers and customers. Heretofore unimaginable, we are on the cusp of creating real-time simulations that not only optimize performance and decision making, but also make decisions, informed by agents representing every possible commercial actor at scale.

One area of Fintech where GenAI is producing mixed results is payments. Last year, a security expert at a big bank told me that, while AI-based fraud monitoring tools are certainly helping to identify fraudulent transactions, they are also producing false positives at an alarming rate, costing nearly as much in orphan transactions as the fraud they are meant to root out. That’s why a Welsh startup, Burbank, caught my eye: Burbank eliminates the need to do most forms of fraud monitoring by turning online payments into card present transactions. The user experience is simple, tap and pin, just like in the store, but the technology concept is complex: payments networks could never withstand 2 billion new terminals, so Burbank invented a way to use the public keys on the users’ phone and derive the keys on demand, obviating the need to tokenize the transaction. Payments was the first vanguard of Fintech. Everything old is new again.

Previously, I’ve written about how AI is powering parametric insurance. The story there is similar to payments. On the plus side, models are getting more sophisticated, allowing insurers to deeply understand complex risk at scale. Meanwhile, the very same models render certain risks uninsurable, creating the potential for vast social and economic problems, especially in the context of climate volatility.

Enter GenAI: startups are already figuring out how to use the same models that predict catastrophe to mitigate risk. In the wake of one of the most devastating wildfires in U.S. history, one GenAI hardware startup focused on wildfire mitigation and prescribed fire operations, BurnBot, is seeing a spike in demand. BurnBot integrates a system of advanced, purpose-built robots, AI, intelligent automation and other proprietary solutions to multiply the scale and impact of conventional (still largely manual) fuel management programs. While advanced algorithmic underwriting may exclude certain properties from traditional insurance markets, AI-powered risk mitigation will ultimately reduce risk, engendering the evolution of emerging risk management solutions.

GenAI’s transformative potential raises questions about the future of work, including the types of human jobs that may be replaced in the name of precision or efficiency. But, in reality, AI isn’t only a problem-solving technology—it is evolving its own interconnected ecosystem of demand, risk and opportunity.

Disclosure: Foxe Capital invested in Burbank in 2024, as a sub-advisor to Anthemis.