The Rise of Agentic AI: How Financial Services Companies Can Prepare

AI is evolving quickly and the next major innovation, agentic AI, is already taking shape. 

Just as most companies are ramping up their generative AI efforts, agentic AI is quickly gaining momentum.  According to a report from Deloitte, a quarter of companies currently using generative AI will experiment with agentic AI pilots this year, and they expect half will do so by 2027. “Some organizations,” Deloitte writes, “will even begin deploying agentic AI into day-to-day workflows by the end of 2025.” 

For most organizations, especially those in highly-regulated industries like financial services, broad deployment of agentic AI is likely a long way off given the significant technical, compliance, and operational effort it requires. Still, it’s important to start building awareness and taking proactive steps to lay the groundwork for this new tool. Early action will not only make future adoption smoother but also strengthen your current AI capabilities. 

Generative AI vs. agentic AI

Generative AI is reactive, requiring prompts to create new content or analyze vast quantities of data to unearth trends or patterns. Although it can learn to personalize its output based on its knowledge of the user, it is still dependent on prompts to provide content. 

Conversely, agentic AI can act and learn independently while leveraging large language models (LLMs), machine learning, and natural language processing (NLP). It works autonomously in an iterative manner to train itself to solve complex problems that require multiple steps to address. 

Agentic AI can make decisions and take action, with little to no human supervision. In fact, agentic AI can plan and execute complex strategies to achieve a goal, gathering data from the external world which it processes and acts on in real-time, such as a self-driving car traveling to a destination across town.

The impact of agentic AI in financial services 

Let’s take financial services as an extended example. Markets move fast, but agentic AI could help firms stay ahead of the curve, analyzing trends and trading patterns to change investment strategies on the fly. AI agents could enable real-time compliance risk assessments as they find and assess anomalies on their own, growing in accuracy as they learn. Agentic AI could also handle mundane, repetitive activities like processing transactions and data entry faster and, ultimately, more accurately, than humans.

In another example, while a generative AI-powered virtual agent can provide human-like answers to customer questions, a chatbot that employs agentic AI can do much, much more. The agentic chatbot could analyze the customer’s investment history, appetite for risk, current financial position, and current sentiment to determine if an opportunity exists to sell an additional financial product. If agentic AI determines it’s appropriate, it can make personalized upsell suggestions. This kind of personalization is critical for remaining competitive—a recent Salesforce report found that 53% of financial customers would switch providers for better digital experiences.

Agentic AI risks and challenges

As enticing as these opportunities may sound, implementing them is still quite a ways off for many enterprises. There are many challenges and risks that need to be addressed before companies can begin implementing agentic AI, especially in a highly regulated environment such as financial services. These risks and challenges include:

Cybersecurity risks

The biggest selling point of agentic AI is that it acts independently, but this independence also introduces security risks. By their nature, agentic AI interacts with a wide array of systems and data resources, both internal and external. Vulnerabilities in the agent’s code or even interference by a malicious actor could result in data leaks, business disruptions, and even monetary losses.

Enabling human oversight

Keeping human beings in the loop will be critical, especially early on. Agentic AI may not be entirely independent for quite some time, and organizations will need to diligently plan how to incorporate human oversight and human-AI hybrid scenarios.

Regulatory compliance 

There’s no AI exception for non-compliance. Any agentic AI application must have guardrails and follow current regulatory frameworks.

Explainability

Often, AI functions like a black box—data goes in, results come out, and no one knows how those conclusions were reached. This lack of transparency is unlikely to be acceptable to regulators. Organizations must take steps to align with emerging governance frameworks and prioritize explainability in their AI systems. 

Laying a strong foundation for agentic AI

While financial services firms may not be able to deploy agentic AI just yet, it’s clearly on the horizon and preparing now is both smart and strategic. The steps you take today won’t just set you up for agentic AI in the future; they’ll also bring immediate value across your organization.

Here’s a brief overview of the role of data preparation and how companies can start preparing for the implementation of AI. For a deeper dive, check out our article: AI readiness: How to lay the groundwork for success.

Start with your data 

As with any AI, agentic AI is only as good as the data it’s built on.

• Audit your data to find gaps, redundancies, and errors. Ensure that your data is standardized with consistent formats and naming conventions. 
• Make sure you have a wide array of data sources available. Accurately tag your data, which will enable models to deliver more accurate insights. 
• Accessible data is key. The cleanest data in the world does no one any good if it can’t be found. Having clean, comprehensive data will not only prepare you for agentic AI, but also improve the output of analytics, business intelligence, generative AI, and many other applications.

Take cybersecurity seriously 

Agentic AI will require strong protections around your data, so start building those safeguards now. Encrypt sensitive information, apply least-privilege access policies, and ensure compliance with regulations like the California Consumer Privacy Act (CCPA) and the Gramm-Leach-Billey Act (GLBA). 

Prepare your organization for what’s next 

The rise of agentic AI is inevitable. The organizations that start preparing today with a strong data and cybersecurity foundation will be best positioned to capitalize on the opportunity when the time is right. Plus, the preparation organizations need to undergo to be ready for agentic AI confer real, immediate benefits for generative AI, analytics, and other applications that rely on sound data and require strong cybersecurity.

At Think Company, we’ve helped many well-known companies in financial services and many other industries future-proof their data and digital infrastructures.  Interested in preparing to leverage new technologies like agentic AI? Let’s connect.