EXCLUSIVE: “Planning For AI” – Adam Lieberman, Finastra in ‘The Fintech Magazine’

Whether it’s buried deep in the operational heart of an organisation or powering customer-facing chatbots, AI is a tool that requires careful preparation and a clear sense of what you want it to achieve, says Adam Lieberman

THE FINTECH MAGAZINE: With so much investment going into AI today, what advice would you give financial services organisations to help them maximise their return on AI initiatives?

ADAM LIEBERMAN: The desire to deliver efficiency and automation with AI is understandably the key driver of adoption, but projects will fail to deliver return on investment (ROI) when there is a lack of solid objectives from the outset. Setting the parameters of success, such as well-defined metrics and KPIs, is crucial.

An organisation with a mature AI roadmap will have clearly defined strategic objectives, against which AI implementation and integration is mapped, so that the technology directly supports organisational goals. Jumping straight into adoption, without considering the technological and cultural limitations and challenges will almost certainly lead to poor ROI. Banks and financial institutions are full of legacy technology, and with that comes a culture that leans towards legacy practices and ways of working.

In this regard, AI projects that support internal workflows, such as those underpinned by generative AI tools, require a shift in the organisation’s mindset that prioritises upskilling employees and updating security and governance frameworks to achieve ROI. For more advanced projects, such as those that relate to the development and delivery of products and services, bottlenecks may occur that can also affect ROI. For example, in industries in which sensitive data are commonplace, gaining access to datasets may take some time or new regulations may emerge that restrict or limit access to data. Being aware of potential restrictions from the outset will help to protect against these bottlenecks

TFM: Chief artificial intelligence officers are still a relatively new addition to the C suite. Speaking as a CAIO yourself, how important is effective leadership when it comes to implementing and developing AI initiatives?

AL: Leaders can be well-intentioned, but if they do not involve all stakeholders from the inception of AI initiatives, successful projects will elude them. Success relies on buy-in and validation of AI use cases from internal stakeholders and delivery teams, from product owners, data science teams and developers, to the C-suite, with all agreeing on KPIs and the wider roadmap. This ensures a critical mass of investment and resource bandwidth and ensures everyone is working towards the same goal.

As well as stakeholder management, AI leaders should be familiar with the core principles and ethical concerns that will govern the development and use of AI-powered solutions across the organisation. Across financial services, there has been a rise in the number of AI-related leadership roles. This is evidence that AI maturity is an increasing priority for the C-suite.

TFM: How should financial services firms measure their level of AI maturity?

AL: AI maturity can be measured against a number of factors, from educational initiatives and skills development programmes, to ensuring the right talent and leadership is in place across technical roles. Measuring how often employees are using the available AI tools is also a good way to assess the general state of enterprise fluency with the technology. When it comes to the development and delivery of AI-powered products and services, maturity is measured by the business’s approach to defining use cases, so that a robust AI roadmap can b determined, incorporating architectural best practices and frameworks, so that systems are able to adapt and scale as the technology advances.

A crucial point for financial services firms is that data infrastructure must be mature enough to ensure organisational data is AI-ready. If this is not in place, the focus must be on data transformation and the modernisation of systems.

TFM: How can organisations upskill their employees and increase AI maturity?

AL: AI maturity will ultimately mean different things for different employees. For example, developer teams will already likely be working with AI coding tools and applications regularly, while marketing teams will still be experimenting with generative AI to assist with content development. There is no one-size-fits-all approach to advancing users from low-maturity to high-maturity, but looking at the requirements of departments and teams across an organisation can help leaders establish bespoke frameworks for upskilling and increasing engagement with AI-powered tools.

Identifying AI evangelists and super users within departments and teams is another way to help advance other employees’ understanding and engagement with available AI tools, enabling them to deliver efficiencies within their own workflows. Creating a collaborative culture around skills development comes back to leadership, as employees must be given the space and opportunity to upskill and share their learnings and expertise with peers.

TFM: What does the future hold for AI in financial services?

“The realisation of fully autonomous agentic AI systems is where we are heading, and banks and fintechs need to understand how they can incorporate them into their service offerings”

AL: AI is poised to supercharge the industry across multiple tiers. Enhanced and streamlined customer support wil be a primary focus, with AI-powered chatbots and virtual assistants providing instant, accurate responses to customer inquiries. These systems will handle complex interactions, from account management to support issues, freeing up human agents to focus on more nuanced issues.

In-product assistance will enable users to navigate traditional financial products through natural language interactions. This will make complex financial tools more accessible and provide a seamless UI/UX experience. Agentic workflows will represent the next evolution, with AI systems connecting to tools and functions, capable of executing complex tasks autonomously, assisting users with their daily tasks. By automating routine tasks and providing data-driven insights, AI will significantly boost productivity for financial institutions, enabling employees to focus on high-value activities.

AI will also continue to drive developer productivity – serving as a first-line assistant to auto-complete code, streamline the documentation process, provide better tes coverage, and overall improve the productivity and enjoyment factor of software development. As AI continues to evolve, its impact will strengthen, creating a future of more efficient, personalised, and intelligent financial services.

TFM: Talk us through some use cases for AI in areas such as payments and lending.

AL: We’re exploring AI to solve some of the real-world challenges facing banks. This includes providing instant assistance to operational staff in areas such as trade finance, lending and payments processing. Trade finance is complex, and it’s notable that the industry faces a significant talent gap. AI tools can help new team members learn and become productive much more quickly through prompt-based assistance. They can navigate processes much more easily without having to sift through extensive documentation.

Payments teams can also benefit in using AI, particularly in relation to data. The power to analyse vast amounts of complex payment data through natural language processing drives more robust insights for banks, particularl around payment activity and in turn about the best products and services to offer.

It’s clear that AI can remove time-consuming, low-value work across the board. Significant efficiency gains can be realised using AI for transcription, for translation and for digitising paper-based contracts. For lending teams, the ability to digitise, query and manage high volumes of complex loan documentation at scale, and to incorporate the data in downstream applications, is transformational.

TFM: What emerging AI technologies should financial services firms be paying close attention to right now?

AL: The rise of AI agents is leading to the development of many business use cases for AI. One of the main reasons behind this is that agents are able to plug into the investment most organisations have made over the last two years in generative AI tools. By extending the capabilities of chatbots, for example, organisations can revolutionise customer experience, as well as deliver advanced knowledge and data search and discovery for internal teams. Through combining agents with frameworks like LangChain, agents can connect LLMs to external data and APIs, allowing them to provide further context and detail when responding to prompts.

The realisation of fully autonomous agentic AI systems is where we are heading, and banks and fintechs need to understand how they can incorporate them into their service offerings. Payments and retail are both areas where use cases for agentic AI systems will be delivered in a meaningful way, with experts believing agentic commerce will account for a significant portion of all commerce by 2030. Connected to this is the use of different protocols for agents and LLMs to connect and communicate with one another. In simple terms, agent-to-agent (A2A) protocols are frameworks that enable them to communicate and collaborate with one another, creating larger, more dynamic AI systems.

The ModelContext Protocol (MCP) is another framework that allows LLMs to access other databases, APIs, and tools, such as agents, extending the capabilities of the LLM to the systems organisations prefer to interact with. These protocols are supporting the rapid evolution of agentic systems and should be key investments for financial services organisations and fintechs alike.

Adam Lieberman will be at Sibos from 29 September to 2 October. If you’d like to speak to the Finastra team at Sibos, visit booth #H036

This article was published in The Fintech Magazine Issue #36, Page 8-9