AI trends in financial services

In this first quarter of 2025, AI advancements are continuing apace. In the realm of financial services, use cases must adhere to restrictions and regulations around data usage, but there are a number of areas that are gaining significant traction and will continue to mature over the next year. Here are six trends that I see gaining traction and delivering value.

1. Increased usage of multimodal models 

Recent advances in multimodal large language models are set to unlock a plethora of opportunities for financial services organizations in the year ahead. As the name suggests, multimodal models are trained to handle data and inputs from multiple sources, such as text, images, video and audio.  

To date we’ve seen financial services organizations focusing more on the text capabilities of Gen AI – instructing models via text-based prompts to access and summarize written information stored in a variety of formats across multiple systems. Going forward, the increased usage of voice will provide additional methods of interaction with models. 

Harnessing the capabilities of multimodal models, and their ability to process different formats of data simultaneously, will be hugely important. Financial services professionals will be able to orchestrate the creation of multimedia presentations and reports – distilling information stored in different formats in systems across the organization into timely and actionable insights. 

A key use case for these models in financial services will be accessing and unlocking the potential of unstructured data that sits in business silos. For example, transcribing customer calls from audio to text, or extracting data from paper-based documents that have been scanned and digitized, to make the data more accessible.  

2. The rise of agentic AI 

Agentic AI is another hot topic for 2025. Agents augment LLMs, going beyond simple information retrieval and performing tasks completely autonomously without human intervention, or with supervision by a human. They can be permissioned to access additional data sources via APIs and to interact with tools and functions to execute specific tasks. So rather than just asking Gen AI to summarize information, we’ll increasingly see LLMs being used to execute actual work. This is going to be a huge trend broadly, but also in financial services.  

3. Growth in deep research 

Deep research AI tools are now becoming available which allow LLMs to search the public internet to conduct thorough, in-depth analysis and research. Such models can synthesize data from thousands of different sources and create detailed reports on specialist or highly technical subjects. In financial services, potential use cases could include detailed research into areas such as ESG investment, a review of the competitive landscape for the provision of a particular financial services product, or an in-depth look into specific regulatory and compliance issues. 

4. AI-powered browser use 

Building on the power of AI agents and multimodal AI capabilities, another area of focus is going to be allowing LLMs to access and control users’ web browsers or systems in order to carry out specific tasks.  A potential use case could include allowing the LLM to assist call center staff by helping them quickly navigate their way to specific support documents or product-help pages to support customers in resolving complex enquiries or technical issues. Over the next year or so, I anticipate we’ll see a lot of tech startups working on solutions for browser use – although of course caution will remain essential when accessing mission-critical financial software.

5. AI-first roles 

Another trend is the rise in job postings that are starting to describe traditional roles in the financial services industry as ‘AI-first’. For example, advertising for an ‘AI-first financial director’. The core competencies of a financial director will still be essential, but in addition, firms are starting to look for candidates accustomed to using Gen AI and no-code solutions as an integral part of their role.

Candidates should be able to demonstrate first-hand experience in using AI to deliver productivity and efficiency gains in their work. Hiring people who can focus on more challenging activities by using AI to automate mundane or routine tasks, is something I expect to grow.  

6. AI-driven software development stacks 

The formulation of AI-driven software development stacks, leveraging code completion tools like GitHub Copilot or Cursor and other supporting tools, will enable forward-thinking organizations to reap huge efficiencies in the software development process.

Furthermore, such technology will help open up and democratize access – enabling non-technical personnel to share their ideas and develop applications that solve specific pain points they’ve identified. When used in conjunction with low-code and no-code platforms, users with fairly limited training can put together reasonably advanced applications, which is another trend I see becoming more prevalent in financial services. 

There’s plenty to be excited about as we consider AI-powered innovation in financial services, and across all industries more broadly. As new capabilities and techniques come online with increasing frequency, there will undoubtedly be developments that unlock all kinds of new use cases over the coming year. As ever, the key to taking advantage of them will be the ability to be agile, experiment and adapt.