After several years positioned as the most transformative technological advancement in financial services, genAI’s new, shinier rival may be poised to steal the spotlight in the next 12 months. FStech news editor Alexandra Leonards explores the opportunities and risks for firms as they explore the latest iteration of artificial intelligence: agentic AI.

The dust hasn’t yet settled on the unprecedented hype around generative AI (genAI) and the financial services industry is already exploring another, more advanced version of artificial intelligence (AI).

While genAI provides responses based on a single interaction using natural language processing, the next level of AI – agentic AI – is able to solve more complex and multi-step problems through sophisticated reasoning and iterative planning. What truly sets the technology apart from previous iterations, its proponents argue, is that it can make decisions autonomously.

GenAI has already begun made its mark in the financial services industry, with firms moving beyond pilots to embed it into their day-to-day operations. If agentic AI follows in a similar trajectory the sector could be completely transformed by the technology by 2027, industry insiders believe.

Widespread optimism

“The emergence of agentic AI is exciting and has the potential to transform customer and colleague experiences across industries, including within financial services,” Rohit Dhawan, director of AI and advanced analytics at Lloyds Banking Group tells FStech. “The technology will create more personalised customer experiences while improving operational efficiency.”

He says that by understanding individual customer needs, behaviours and preferences, agentic AI will be able to handle tasks on their behalf while providing helpful financial guidance.

“For example, agentic AI would be able to analyse customer data and behaviour to provide personalised product recommendations,” continues Dhawan.

Agents powered by agentic AI will be able to support colleagues by automating routine administrative tasks, including everything from managing compliance checks to processing documents, freeing up colleagues to focus on more complex and strategic tasks.

Dhawan says that while the technology is still emerging, its potential is vast, with the industry likely to see applications in areas such as customer interactions, fraud detection, risk management, and compliance.

Jürgen Eckel, managing director and partner at global management consulting firm Boston Consulting Group (BCG) predicts that there will be more experimentation with the technology this year, agreeing that the key opportunities for agentic AI lie in compliance, client onboarding, and fraud detection, where it can enhance efficiency and augment human expertise.

“Many firms are already exploring Agentic AI,” he says. “BCG research found that there’s widespread optimism about the potential of AI agents, with 67 per cent of companies considering them in their transformation journeys.

“With the potential to deliver up to three times more productivity and speed, AI agents are a top priority for business leaders in 2025.”

BCG expects to see the first real-world implementations in customer services across roles like contact centre agents, supervisors, or domain experts, as well as in research and capital market environments.

Lloyds itself is well on this journey, with the bank currently developing a range of solutions that use agentic AI to support both its employees and customers, with more details about the technology to be shared in due course.

“The shift is moving beyond personal productivity tools, towards deeper process integration,” adds Eckel. “We are seeing financial institutions explore the use of foundation models in investment banking and wealth management.”

Fouzi Husaini, chief technology & AI officer at the “world’s first modern card issuing platform” Marqeta, says that the most transformative use cases the industry is likely to see in the more distant future will be oriented around personalisation.

“The days of the status quo are numbered as consumers crave tailor-made financial experiences,” he explains. “Agentic AI can allow FinTechs to provide personalised financial services that help consumers and businesses make their money work better for them.”

This could involve analysing consumer spending data, allowing for tailored financial offers and services.

“We could also see AI ‘predictive cards’, which can anticipate a consumer’s spending requirements based on their past behaviour,” continues Husaini. “This allows credit limits to be adjusted and tailored rewards to be offered automatically, based on patterns in their past behaviours, creating a personalised experience for each individual.”

The technology chief also describes Agentic AI’s impact on fraud detection as a potential “game changer”, and believes many financial providers are likely already working to integrate it into their fraud protection offerings.

“While human agents may experience decision fatigue or be slow to respond, agentic AI can be constantly vigilant for data indicating potential cases of fraud,” he adds. “This is vital in a space where one missed signal can lead to significant consequences.”

The technology’s ability to learn and adapt also promises to consistently develop its fraud detecting skills, meaning it will improve in this task as scams evolve over time.

Required guardrails

“Agentic AI could significantly enhance customer and colleague experiences, but it is important that organisations carry out robust testing and ensure the right safeguards are in place before making the technology available,” says Lloyds Banking Group’s Rohit Dhawan.

The bank, like many across the sector, has established a specific AI Centre of Excellence which acts as research and development hub. Dhawan says the centre works to make sure that Lloyds remains at the cutting edge of AI technology, bringing together experts including Magdalena Lis, the group’s head of responsible AI who started work at the beginning of February. Beyond Lis, Lloyds boasts that the unit has expanded rapidly to over 200 experts in the field who collectively hold over 30 PhDs in key disciplines.

“We are working at pace to harness the benefits of AI, whilst also carrying out testing and development to ensure any new propositions are safe before being made available to colleagues or customers,” continues Dhawan.

Some of the key risks of the technology in financial services include cybersecurity threats from vast data reliance, potential market volatility from synchonised decisions, and conflicts between autonomous agents and human oversight.

“First, navigating agentic AI’s dependence on vast quantities of data could present significant cybersecurity risks if it’s not implemented correctly with required guardrails,” says BCG’s Jürgen Eckel. “Second, agentic AI has the potential to drive abrupt market dynamics, so there could be a broader systemic risk of market volatility from its synchronised decisions.”

He warns that the autonomous capabilities of these agents may conflict with human interests and outpace traditional security measures.

Research from BCG has also shown a growing demand for upskilling, which presents as much room for positivity as it does concern.

Eckel says that AI training equally poses a risk to enterprise governance as employees will increasingly have the toolset to introduce inconsistencies within these automations.

Mauricio Toro-Bermudez, data scientist at savings and money transfer app Cheddar, says that agentic AI could give misleading or harmful advice, leading customers to lose money or make poor decisions.

“If that happens, the company could face legal consequences, especially if they didn’t even know the AI provided the advice,” he adds.

He also explains that while personalisation can be valuable, it must be done carefully.

“For example, Netflix’s race-based predictions for TV show recommendations caused an uproar and had serious reputation consequences,” continues Toro-Bermudez. “In finance, similar mistakes could be even worse, especially when predicting things like a person’s pregnancy or home buying status.”

Firms must prioritise privacy and ensure that they get user permissions before using bank transaction data for AI-driven decisions so that personalisation doesn’t cross any lines — both ethically, and legally in light of the stringent requirements of the EU AI Act and GDPR. Particularly, as Eckel points out, because agentic AI demands far greater safeguarding of consumer privacy and data.

“Companies will need to increase their efforts to educate consumers on the associated risks of agentic AI,” he warns. “From a technology perspective, firms will need to implement stricter best practices on data sharing and encryption.”

Ultimately, monitoring the output of AI agents will be an important early step to assess decision-making biases and ensure overall consumer satisfaction.

Some of the risks associated with genAI also apply to the latest iteration of AI. For example, agentic AI also has the potential to generate factually incorrect information or ‘hallucinate’.

But Marqeta’s Fouzi Husaini says that the technology is much less prone to this type of error than genAI because of its enhanced ability to learn.

“Agentic AI is designed to operate within clearly defined parameters and follow laid-out workflows, meaning it has built in guardrails to significantly reduce the chance of any errors,” he adds.

Jobs safe… for now

Concern about technology taking over jobs is a tale as old as time. But in financial services, the introduction of genAI has largely been embraced as a method for enhancing employee experience, rather than a mode to replace workforces.

However, given that the technology takes AI a step further by facilitating autonomous decision-making—something that could only previously be done by a human—can the same really be true of agentic AI?

Cheddar data scientist Toro-Bermudez says that like in other industries, there is a risk the tech could replace certain customer service jobs in the financial space.

“AI chatbots can handle many inquiries, so fewer humans might be needed for basic tasks,” he continues. “However, more specialised jobs, like financial advice, are still safe…for now.”
By 2026, he predicts that the industry could see heightened levels of automation, even in more specialised areas. With this in mind, he says it’s important to “keep an eye on the future of work”.

Eckel believes that from an employee perspective, the biggest challenge will be reimagining workflows and driving cultural and organisational change.

“Our research shows that executives increasingly view AI and talent as complementary, with only seven per cent of leaders expecting headcount reductions due to AI automation,” he continues. “However, successful AI integration hinges on upskilling—yet only one-third of companies have trained a quarter of their workforce in AI, highlighting a gap that must be addressed for employees to feel confident using this technology.”

He says that a lot of financial services players are looking to agentic AI in the context of efficiencies, which means the sector can expect a “short-term” reduction of demand for white-collar work.

“Long-term effects are close to impossible to predict, but supervising, creating, training and maintaining agentic setups will create new classes of jobs,” explains Eckel.

Barriers to integration

As the industry saw with genAI, there is still hesitation from some firms to take the leap with the latest iteration of AI technology.

One key barrier to integration will likely be balancing agentic AI’s enhanced capabilities with human supervision.

“Agentic AI can carry out complex, multi-step tasks without human supervision and can learn and adapt over time,” says Husaini. “This means it has the ability to act autonomously, and it is vital that this is met with the correct amount of human oversight.”

He continues: “It will be important to balance letting the technology adapt and react on its own with oversight, which risks stifling the technology’s capabilities if too pervasive. However, as agentic AI is engineered to act within clear, defined parameters, it will not be a difficult problem to solve.”

Compliance concerns will continue to be a major area of concern in the highly regulated financial services space, with financial institutions walking a tightrope of balancing needs for explainability and transparency with consumer protection all while attempting to continue innovating.

“Regulations are strict, so using AI for more complex tasks is tricky,” says Toro-Bermudez. “While customer service AI has been successfully used in other industries, financial advice still faces more hurdles due to the sensitive nature of the information and the need to protect customers from bad recommendations.”

Eckel agrees that regulation will be a challenge for those looking to adopt the technology.

“Regulators’ perception or misconception of ‘uncontrolled computer systems taking decisions’ as well as the public perception will likely be initial hurdles,” he says. “Any implementation requires a thorough assessment of cyber and implementation risks, proper guardrails, as well as a comprehensive responsible AI process and design.”

But for BCG, the most obvious barrier is internal resistance within organisations, likely due to employees’ fears around displacement or their struggle to adapt to new workflows, lack of expertise, and ethical and regulatory concerns.

In the early days of genAI, there was much trepidation about how and where the technology should be adopted in financial services, with widespread concerns about regulation, ethics and jobs, as well as uncertainty about which use cases would be most valuable to operations.

However, it did not take long before firms moved beyond the trial stage to implement the technology in their everyday operations, with financial services providers continuing to identify new and improved ways to boost internal and customer-facing processes.

Much like genAI, agentic AI is poised to redefine financial services, offering significant enhancements in risk management, compliance and customer personalisation. However, successful adoption will depend not only on firms’ willingness to integrate the technology but also on their ability to establish strong governance frameworks, ensure ethical AI use, and navigate regulatory complexities

With recent AI advancements still fresh in the minds of its decision-makers, financial services firms have an opportunity to lead the charge in responsible and effective implementation of agentic AI.

---