Join

Getting Started with AWS Generative AI in Financial Services

AWS has built one of the most comprehensive free training ecosystems for generative AI anywhere in the world. Whether you are completely new to AI or ready to build your first application, there is a pathway for you, and most of it costs nothing.

Generative AI is moving fast and the skills gap is real. A study by AWS and Access Partnership found that hiring AI-skilled talent is a priority for 73% of employers, yet three out of four say they cannot meet that need. At the same time, workers with AI skills can earn up to 47% more in salary. The opportunity is significant, and AWS has invested heavily to help close that gap, committing to train 2 million people globally through its AI Ready initiative.

This article covers three things: where to start learning, what AWS Prescriptive Guidance for generative AI means in practice, and how the AWS Well-Architected Generative AI Lens can help you build AI systems the right way.

Why financial services is a priority for generative AI

Financial institutions have always sat on vast amounts of data. Customer records, transaction histories, regulatory filings, loan applications, claims documents, risk reports. For decades, most of that data was locked away in systems that required expensive specialists to access and interpret.

Generative AI changes that. It makes it possible for analysts, compliance teams, customer service staff, and product managers to interrogate that data, surface insights, and automate repetitive tasks without needing to write code or run complex queries.

According to a Moody's study cited in AWS research on agentic AI in financial services, 70 percent of financial institutions surveyed now prioritise AI for risk and compliance work. Another 66 percent are using it to accelerate analysis, and 64 percent to reduce costs. These are not long term ambitions. They are things happening now, at institutions of every size.

The applications that are seeing the most traction include fraud detection, regulatory document review, customer service automation, personalised communications, and internal productivity tools. Mastercard uses AWS AI and machine learning services to detect and prevent fraud while keeping customer journeys as smooth as possible. NatWest Group runs nearly 100 machine learning models through AWS to personalise how it serves customers.

What is consistent across these examples is that the value is not coming from one big transformation project. It is coming from solving specific, costly problems faster than was previously possible.

Where AWS fits in

AWS is the cloud platform that the majority of large financial institutions already use, which makes it a natural starting point for building with generative AI. It offers the infrastructure, the security controls, and the compliance frameworks that financial services organisations need. It also offers a growing set of tools designed to make generative AI accessible, from no code experimentation environments through to enterprise production deployments.

The core service for building generative AI applications on AWS is Amazon Bedrock. It gives access to a range of leading foundation models, including models from Anthropic, Meta, Mistral, and Amazon itself, through a single API. Crucially for financial services, data sent to Bedrock is not used to train the underlying models and stays within your AWS environment. That matters in a sector where data privacy and regulatory obligations are non negotiable.

For teams that want to work with AI without writing code at all, Amazon SageMaker Canvas allows users to build and run machine learning models through a visual interface. And for organisations that want a ready made AI assistant connected to their internal data, Amazon Q Business can be deployed on top of existing document stores, wikis, and data sources within weeks.

What AWS prescriptive guidance says about building for financial services

AWS publishes prescriptive guidance for organisations that want to move beyond proof of concept and build generative AI into production systems. The most relevant document for financial services teams is Building an Enterprise Ready Generative AI Platform on AWS, which lays out a four layer approach covering infrastructure, model selection, security and governance, and reusable application patterns.

The governance layer is particularly important in financial services. The guidance specifically addresses how to build access controls around foundation models, how to implement audit logging for AI outputs, and how to create approval workflows that keep human oversight in the loop. These are not optional extras in a regulated environment. They are the foundations that make everything else defensible to a regulator, a board, or an audit team.

The prescriptive guidance also addresses one of the most common barriers to scaling AI in financial services, which is the tendency for teams to build isolated pilots that cannot be reused. By establishing shared platforms and reusable patterns, organisations can move from running one AI experiment to running twenty, without proportionally increasing the risk or the overhead.

Worth reading: The AWS financial services resources page at aws.amazon.com/financial-services/resources contains ebooks, whitepapers, and recorded sessions from institutions including Nasdaq and Sun Life. It is a useful reference if you want to see how peers are approaching specific problems before building your own solution.

The Well Architected Generative AI Lens and what it means for your team

The AWS Well Architected Generative AI Lens is a framework for assessing whether an AI workload has been designed soundly. It covers six pillars: operational excellence, security, reliability, performance efficiency, cost optimisation, and sustainability.

In financial services, three of these pillars tend to require the most attention.

Security covers not just data protection but model access controls, prompt injection risks, and output monitoring. A poorly governed AI system in a bank is not just a technical problem. It is a conduct risk, a regulatory exposure, and potentially a reputational issue.

Reliability covers what happens when a model behaves unexpectedly or a system goes down. Financial services organisations need AI systems that fail gracefully, that have fallback processes, and that are monitored continuously rather than checked occasionally.

Cost optimisation matters more than many teams realise at the outset. Inference costs can escalate quickly if model selection and usage patterns are not managed deliberately. The Lens provides specific guidance on techniques like batching requests and using smaller models for preprocessing tasks, both of which can reduce costs significantly at scale.

AWS also launched a dedicated Responsible AI Lens in late 2025, which sits alongside the Generative AI Lens and provides structured guidance on bias, fairness, transparency, and governance. For organisations navigating frameworks like the EU AI Act or FCA guidance on AI in financial services, this provides a practical structure for evidencing how AI systems have been designed and monitored.

Free AWS training to build your knowledge now

One of the most useful things about the AWS training ecosystem is that most of it is free and self paced. You do not need to be in a technical role to benefit from it, and you do not need to wait for your organisation to sponsor a course.

Here are the most relevant starting points for people in financial services:

Where to begin: Go to skillbuilder.aws, create a free account, and start with the Introduction to Generative AI. Once you have that foundation, the Prompt Engineering course will give you something practical you can apply to your work the same week.

Further reading

Author

Rossana Bianchi is Head of Generative AI Strategy at the AWS Generative AI Innovation Center for Australia and New Zealand, where she works with organisations across the region to shape and accelerate their generative AI programmes. She brings over a decade of consulting experience working with some of the world’s largest organisations on AI strategy, governance, and responsible innovation. Prior to AWS, she led the AI and Data Ethics Capability at KPMG, advising on AI risk management and governance across financial services and beyond. Rossana works closely with academia and policymakers on practical approaches to ethical AI, and has written on the subject for Tech For Good. She holds an executive certificate in Artificial Intelligence and Business Strategy from MIT, and completed the FinTech Executive Programme at the University of Hong Kong. She has been nominated in the UK for diversity and inclusion awards as a game changer in technology. Rossana is a member of the RepresentAI community and is passionate about making AI accessible, equitable, and trustworthy for everyone. Connect with her on LinkedIn.