News

Four AWS Tips for Agentic AI in Financial Services

• By David Ramel
• 09/09/2025

Amazon Web Services (AWS) is offering new guidance to financial institutions exploring the use of agentic AI.

AWS's message is clear: financial organizations see agentic AI as a strategic necessity, but successful implementation requires careful planning. While some firms already have dozens of agentic systems in production, many face challenges related to skills, security, and legacy technology. To help, AWS outlined four best practices that institutions can follow as they move from experimentation to deployment.

The company published the recommendations on Sept. 8 in a blog post authored by Andrew Renzella, AWS's global financial services industry analyst lead. The post draws on a Forrester study commissioned by AWS that surveyed more than 550 global technology and business strategy leaders.

"The financial services industry is leveraging AI to transform how financial institutions serve their customers," Renzella said. "AI solutions can help proactively manage portfolios, automatically refinance mortgages when rates decrease, and negotiate insurance premiums for customers."

Here are summaries of the four AWS recommendations for agentic AI success.

1. Start Internally With Low-Risk Use Cases
AWS recommends launching agentic AI in employee-facing scenarios before exposing the technology to clients. This lets organizations validate systems in controlled environments. One executive explained, "We really don't want to do anything client-facing right now. But as we start to see the value in these employee-facing use cases and we start to see the true ROI, we're going to start opening it up more to clients."

2. Invest in Training and Change Management
AWS cautions that agentic AI is not just a technology rollout but an organizational shift. "The change management part is very, very critical. Part of change management is the education and training. This is new technology, and especially for people that aren't that tech-savvy, they don't understand what it is or all the capabilities that it could bring," said one SVP. Training both technical and non-technical staff is necessary to demystify AI systems and drive adoption.

3. Build Strong Partnerships
Third-party providers play a key role in agentic AI strategies. The Forrester survey found that 84% of financial services organizations depend on partner integrations. AWS suggests selecting partners that combine technical expertise with a deep understanding of financial regulations and security.

4. Design With Security From the Start
Security frameworks must be built in from the beginning. As one SVP noted, "If there's a breach, it could have a significant impact on the company, the resources, and our overall reputation. The security, the risk piece, that's been the biggest challenge to work through". AWS recommends comprehensive controls around data, models, access, and monitoring.

The Three Pillars of Autonomous Finance
AWS also discussed the three pillars of autonomous finance identified in the Forrester study, describing how agentic AI is already reshaping financial services. Here they are, summarized:

1. Customer service with AI: Institutions are exploring agentic AI for banking services such as account management, loan applications, and dispute resolution. These systems are designed to complement existing customer service teams by speeding up processes and making them more efficient.
2. AI in financial operations: Behind the scenes, agentic AI is being used to optimize workflows, analyze market conditions, and adjust risk parameters in real time. Unlike traditional automation, these systems make intelligent decisions that adapt to changing data.
3. Hyper-personalized financial guidance: Organizations expect agentic AI to democratize financial advice, offering tailored guidance on mortgages, investments, and risk management at scale. What was once reserved for high-net-worth individuals may soon become broadly available.

Agentic AI in Finance
Agentic AI differs from traditional generative AI by autonomously perceiving, deciding, and acting while learning over time. In financial services, that means systems that can manage portfolios, process loans, and provide personalized advice without direct human prompting. According to AWS, some organizations already run agentic systems in production and plan to expand usage significantly by 2026.

AWS Tools Supporting Deployment
Naturally, AWS provides a technology stack designed for these use cases, including Amazon SageMaker for machine learning model development, Amazon Bedrock for secure access to foundation models, and Amazon Bedrock AgentCore for building and managing autonomous agents. The AWS Marketplace also offers pre-built agents, professional services, and integration tools.