Ashish Chopra is CIO at TDECU.
Agentic artificial intelligence (agentic AI) is ushering in a new era for financial institutions, offering transformative capabilities that can fundamentally reshape operations, improve customer engagement and enhance risk management.
By harnessing the power of agentic AI, financial organizations can automate complex, end-to-end tasks, achieve goal-oriented reasoning, unlock value from diverse data and provide hyper-personalized intelligence. These capabilities position financial institutions to realize new levels of efficiency, accuracy and innovation, underpinned by data-driven decision making.
In this article, I’ll look at a few of the key tenets that make agentic AI so effective, as well as a few challenges I’ve identified that must be solved before its full adoption and success.
The Hyper-Personalization Behind Agentic AI
The effectiveness of agentic AI lies in its ability to combine multiple AI techniques and advanced technologies. Key components include:
• Machine Learning (ML): At the foundation of agentic AI, machine learning enables these systems to learn from vast datasets, continuously improving their performance and decision making accuracy over time.
• Reinforcement Learning (RL): Through reinforcement learning, agentic AI learns from its interactions within its environment and improves based on feedback. This capacity allows the system to optimize its actions and strategies in real time.
Types Of AI Agents In Financial Institutions
The complexity and autonomy of agentic AI can be further understood by examining the different types of AI agents, each designed for specific tasks and environments.
Below, I’ll highlight various AI agents and their applications within the financial sector:
1. Reactive Agents (Fixed Automation): These agents respond to immediate stimuli based on predefined rules or conditions without memory of past events. They are suitable for simple, repetitive tasks, such as flagging transactions that meet predefined fraud detection criteria.
2. Single-Agent Systems: These systems feature a single AI agent equipped with tools to handle specific tasks. They operate autonomously to perform step-by-step processes based on reasoning and pre-determined actions, such as automating the entire insurance claims process.
3. Model-Based Reflex Agents: These agents maintain an internal model of the world based on their perception and memory of past states. They can adapt to changes in the environment and make decisions using updated information, such as assisting in loan processing by keeping track of applicant profiles and current underwriting rules.
4. Multi-Agent Systems With A Shared Goal: Multiple agents collaborate toward achieving a common objective, maximizing a utility function to optimize outcomes across different solutions, such as portfolio management systems that analyze risk, return and diversification to optimize investment strategies.
5. Multi-Agent Learning Systems: These agents continuously improve their behavior by learning from their interactions and experiences. Often using machine learning techniques, they predict future trends based on past data, such as hyper-personalization engines in banking apps that bundle products and offer unique pricing based on user history and preferences.
6. Complex Multi-Agent Systems (CMAS): These systems involve multiple independent agents, each specializing in different tasks but working together to tackle complex problems. The agents may collaborate using various tools, APIs and data sources, such as an AI-powered system that helps customer service representatives by providing real-time account information and next-best offers and suggesting potential solutions to customer inquiries.
Key Challenges In AI Agent Implementation
Implementing AI agents in financial institutions comes with several challenges:
• Regulatory Compliance And Governance: There is a risk of regulatory fines, penalties and reputational damage if AI systems cannot meet transparency and oversight standards. For example, a fully autonomous AI that approves loans or flags suspicious transactions might not easily provide the level of transparency regulators require.
• Data Privacy And Security: There is an increased risk of data breaches, legal consequences and loss of customer trust if sensitive data is not properly protected. If an AI system is hacked or targeted in a cyberattack, sensitive customer information could be exposed.
• Explainability And Transparency: Challenges in providing clear explanations for AI decisions can lead to legal issues and customer dissatisfaction. For example, if an AI system denies a loan application, the bank might struggle to provide a clear explanation for the decision.
• Ethical And Algorithmic Bias: Potential legal action, reputational damage and regulatory scrutiny over biased AI decision making can affect fairness. An AI system trained on biased data might disproportionately reject loan applications from certain demographic groups.
• Development Issues: Poorly defined prompts and evaluation challenges lead to inefficient systems, slower processes and missed opportunities, impacting customer satisfaction and operational efficiency. For example, if a bank’s AI is asked to “process loan applications,” but the prompt isn’t detailed enough, it may miss crucial information like credit scores or application status.
• LLM Issues: The difficulty in steering AI, high operational costs, reasoning failures and planning issues could result in fraud detection failures, wrong decision making and higher costs in running the system. For example, if an AI is assessing a loan application and the applicant has a non-standard income source, the system might fail to reason properly about this and automatically reject the application.
• Production Issues: Without proper guardrails, scaling solutions, fault tolerance and mechanisms to prevent infinite loops, banks could face compliance violations, customer dissatisfaction, system downtimes and financial losses. For example, if an AI agent repeatedly asks for the same verification information when a customer tries to log in, it may frustrate the user and lead to system failure.
The Future Of Banking
Looking ahead, the integration of agentic AI, real-time payments and blockchain represents a seismic shift in the banking landscape. These technologies are not just supplementary tools; they are reshaping the core operations of financial institutions. The financial world is on the brink of a new era marked by greater efficiency, innovation and customer-centric services.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?