Autonomous AI agents: outsourcing decision-making as the new CX paradigm in finance

Written by Dave Wallace
11th October 2024

My word of the week is “agentification”. Why? Because it’s everywhere. Something must be going on.

AI agents are poised to become integral to how we manage and interact with our finances

The concept of agentification refers to the rise of AI agents, a pivotal development that could redefine the future of financial services. Throughout the evolution of computing, we’ve progressively outsourced various tasks to machines:

Computational tasks: Early computers handled calculations that were laborious for humans.
Knowledge storage and access: The internet and search engines revolutionised how we store and retrieve information.
Content creation: Generative AI (GenAI) has enabled machines to create content, from writing articles to generating art.

Now, with AI agents, we’re on the cusp of a new era – outsourcing decision-making.

Big Tech companies have recently been busy announcing plans around their agent propositions. OpenAI has just completed a $6.5 billion investment round, valuing the business at $150 billion. One of its key new developments has been its “agentic” plans. According to Kevin Weil, OpenAI’s chief product officer: “We want to make it possible to interact with AI in all of the ways that you interact with another human being. These more agentic systems are going to become possible, and 2025 is going to be the year that agentic systems finally hit the mainstream.”

While we may be entering an AI agent hype cycle, the potential of these technologies is too compelling for them not to succeed in the long term. The building blocks are nearly in place.

AI agents are digital entities capable of performing tasks with a degree of autonomy on behalf of a user or another program. These tasks can range from gathering information to making decisions or interacting with other systems and agents, often without direct human intervention.

As AI agents become more intelligent and autonomous, we may soon enter a world where software is aware and capable enough to make decisions on our behalf. The implications are profound, especially for financial services—a sector heavily reliant on decision-making.

It could even become the predominant customer experience paradigm, eventually making the internet and mobile banking obsolete.

Some financial services companies are already taking tentative steps towards agents for customer applications.

Basic algorithms have been the ‘brains’ behind some of the chatbots a few banks have launched. However, their rudimentary nature has often missed the mark from a customer perspective, highlighting the urgent need for more advanced AI solutions.

GenAI has ushered in new possibilities with prompt-based interfaces and plain language responses. Klarna, the BNPL company, has demonstrated the power of automated intelligence by using GenAI to power its new chatbot. Powered by OpenAI, it achieved remarkable results one month after going live:

It had 2.3 million conversations, two-thirds of Klarna’s customer service chats.
It did the equivalent work of 700 full-time agents.
It was on par with human agents regarding customer satisfaction score.
Its problem resolution accuracy led to a 25% drop in repeat inquiries.
It has enabled customers to resolve problems in less than two minutes compared to 11 minutes previously.

Klarna demonstrates that GenAI is the key to unlocking the potential of autonomous agents, hence OpenAI’s interest. According to McKinsey in its recent report, “Why agents are the next frontier of generative AI”, when agentic systems are built using foundation models – which have been trained on extremely large and diverse unstructured datasets – rather than predefined rules, they gain the ability to adapt to different scenarios. This adaptability mirrors how large language models (LLMs) can intelligently respond to prompts they haven’t been explicitly trained on.

With AI agents advancing, we can envision several new possible use cases, including the ability to:

Analyse spending, make recommendations on actions and then perform those actions. For instance, cancelling subscriptions.
Analyse markets, make recommendations on investments and then execute those recommendations.
Optimise opportunities for rewards. I recently wrote an article about account switching. In the UK, generous rewards are offered for switching. An agent might look at the best options and initiate the switch.

Huge implications need to be considered and addressed as we barrel towards this agent-centric world. The transition from predictable outcomes based on coded rules to uncertain results driven by data and intelligence is a significant shift. Regulators will want protection for customers. So, ensuring AI agents comply with all relevant laws and regulations is crucial to avoid legal pitfalls and protect consumers.

There are also ethical considerations. Delegating decision-making to AI raises questions about accountability, fairness and transparency. Who is responsible if an AI agent makes a poor decision? AI agents will also need access to sensitive financial data, so security will be paramount. Finally, customers will need to trust AI agents to make decisions on their behalf. Building this trust will require openness, reliability and demonstrable benefits.

I strongly believe in an agentic future. From my studies in psychology, I’ve learned that the brain likes to conserve energy – it seeks the line of least resistance. Much of human progress has focused on making our lives easier, reducing effort while maximising outcomes.

Technology has continually taken over tasks that require mental effort, effectively outsourcing aspects of our cognition. Outsourcing decision-making to AI agents is a natural progression in this trajectory. It’s a no-brainer (pun intended).

Agentification represents a significant leap forward in the evolution of AI and its application in financial services. While challenges remain, the benefits of increased efficiency, personalised services and the ability to handle complex tasks make AI agents a tantalising prospect for the future.

As technology matures and companies navigate the associated ethical and regulatory landscapes, AI agents are poised to become integral to how we manage and interact with our finances.

About the author

Dave Wallace is a user experience and marketing professional who has spent the last 30 years helping financial services companies design, launch and evolve digital customer experiences.