Fintech applications for AI Agents
Note: Over the past five months, I haven’t published any posts because I became a dad and was on paternity leave. Starting this month, I’ll be back to posting more regularly.
Artificial Intelligence is by far the biggest tech trend happening right now. Everyone in tech is talking about it, VCs are pouring billions of dollars into AI startups, AI has become a geopolitical issue, and not a single day passes without a major announcement.
Within the vast realm of AI, AI Agents are taking a prominent position, increasingly described as the inevitable evolution of SaaS and the future of work.
The goal of this post is to explore how the fintech scene is adapting (or not) to the advent and large scale propagation of AI Agents and who is building in that direction. Plus my own thinking on what key features will be necessary for these new “customers.”
What is an AI Agent?
An AI Agent is a software system capable of autonomously perceiving its environment, processing information, making decisions, and executing actions to achieve a goal. Unlike traditional rule-based automation, AI Agents leverage Large Language Models (LLMs), APIs, and external tools to operate with greater flexibility and reasoning.
In fintech, AI Agents can transform financial applications by reducing manual workflows, enhancing decision-making, and enabling intelligent automation. These agents can ingest and analyze vast amounts of structured and unstructured financial data, interact with APIs, and provide contextual recommendations or actions—without requiring human intervention.
Instead of merely executing pre-defined scripts, AI Agents continuously adapt to new inputs, optimize financial operations, and unlock new ways of interacting with financial applications.
What could they do?
AI Agents in fintech can be seen as intelligent synthetic assistants that go beyond chat-based interactions—they can automate entire financial workflows, monitor real-time data feeds, and execute complex decision-making processes, all while integrating with banking, trading, and compliance infrastructure.
Essentially, every UI could be replaced by an agent that performs tasks on behalf of the user. As a result, many white-collar jobs could theoretically be replaced by AI Agents.
Financial Applications for Agentic Workflows
If we assume that a significant portion of future economic activity will be mediated by synthetic workers (AI Agents), then it becomes crucial to explore how financial applications will need to evolve to accommodate non-human users, aka AI Agents.
Agentic Workflows
First, financial applications will need to adapt their approach for synthetic workers. This means financial use cases must be redesigned to integrate seamlessly into agentic workflows. From my perspective, this primarily involves three key actions:
Supporting the most common agentic frameworks—such as Vercel, LangChain, CrewAI, and AutoGen—since these platforms will serve as the foundation for AI Agent development.
Customizing interfaces (primarily programmatic interfaces—APIs) to enable Agents to execute workflows efficiently.
Optimizing API responses to ensure they are easily digestible by non-deterministic agents. In simple terms, API responses should be streamlined to include only the necessary data, allowing agents to focus on the key values required to complete their workflows.
A prime example of this new interfacing paradigm is Stripe’s Agent Toolkit. This toolkit—which currently supports Vercel, LangChain, and CrewAI—enables financial tasks to be embedded into agentic functions, facilitating complex workflows such as creating a customer, generating a payment link, issuing an invoice, and more.
The toolkit is specifically optimized for agentic workflows: Stripe has minimized the SDK’s footprint and refined API request and response bodies to help agents focus on the essential values they need to process.
Human Control
Allowing AI Agents to perform financial tasks isn’t enough to enable the widespread adoption of agentic financial applications.
The biggest challenge is building user trust—getting people comfortable with delegating financial decisions to AI. This will take time, just as it took time for people to trust websites with their credit card data.
To facilitate this transition and provide users with a sense of control, new solutions are emerging that introduce human oversight in agentic workflows. One notable example is HumanLayer, which enables human approval as part of an AI-driven financial process.
Authentication
The final major challenge that must be addressed to drive substantial adoption of agentic financial workflows is authentication.
If users are expected to trust AI Agents with complex financial tasks, these agents must be able to handle authentication securely and seamlessly. This issue is somewhat mitigated for websites offering API-based authentication, but it remains a significant hurdle for platforms that do not.
What’s needed is an AI Agent-centric identity layer for the web. One interesting initiative in this space is Anon, a service that enables developers to authenticate agents and allow them to act on behalf of users across popular websites.
Conclusions
The tech world is on the cusp of a major shift that will redefine how people interact with the web. What we currently see as AI applications—such as ChatGPT—barely scratch the surface of the impact these technologies will likely have.
Specifically, AI Agents will fundamentally change how people conduct business and interact with each other, with financial transactions playing a key role in this transformation.
Existing fintech players built their success by creating solutions for digital users. However, the current paradigm is not easily extendable to synthetic workers. AI Agents will require their own playbook and purpose-built services.
Just as fintech startups in 2010 built for the mobile era, fintech startups in 2025 must build for the Agentic era. Ignore it at your peril.
Resources
AI Agents
Issues with flows
FInancial applications for AI Agents
https://www.fintechbrainfood.com/p/state-ai-financial-services
https://www.anon.com/
Agentic economy