Financial services is entering a new chapter, according to Telefónica Tech, as agentic AI and hybrid human-agent operating models begin to reshape how institutions think about work, customer service and operational design.

In the third edition of its Human Side of Finance series, titled Taking Financial Services to the Frontier, created by Telefónica Tech’s head of banking Stephen O’Keane, he argues that the sector is moving beyond digitisation and basic automation towards what he calls the ‘Frontier Firm’ model – a framework in which humans and AI agents work together, but people remain central to judgement, empathy and accountability.

There’s plenty of data out there to support what O’Keane posits: Research from AccountsIQ found that organisations favour solutions that provide transparency, clear audit trails and human approval controls.

Beyond this, a recent report from Scottish fintech Aveni found that the use of AI in financial services faces a ‘reality check’, with firms facing major gaps in critical areas like governance, data quality, and cultural readiness.

The message is clear: the next stage of AI adoption in financial services is not simply about deploying new tools more quickly, but about redesigning processes so colleagues can spend less time on repetitive administration and more time on higher-value decision-making.

How did we get here?

Telefónica Tech sets out a simple progression.

First came digitisation, which moved paper-based and manual processes into digital channels and systems. That improved access and reduced friction, but largely left the underlying process unchanged.

Then came automation, where rule-based workflows were streamlined to reduce handoffs, lower cost and remove repetitive effort. Yet automation still depends on predefined logic, meaning it tends to speed up the old process rather than fundamentally transform it.

The real shift, O’Keane says, comes with agentic AI – a model in which AI agents do more than execute tasks. In this version of the operating model, agents can collaborate with teams, adapt to context and orchestrate multi-step workflows at scale.

The company describes three patterns within this evolution: Human with Assistant; Human-Led Agents; and Human-Led, Agent-Operated operations, where agents can run entire processes end to end but still escalate to humans when judgement, empathy or complexity demand it.

It’s a notion that O’Keane pithily surmises in the ebook, referencing an internal presentation at IBM from back in 1979 that states: ‘A computer can never be held accountable… Therefore a computer must never make a management decision.’

For financial services, the implication is that AI should not be seen as a replacement for people, but as a way of redesigning work. Administrators become advisers, process executors become decision-makers, and operating the system gives way to shaping outcomes.

Why the sector feels the pressure now

O’Keane posits that the urgency comes from both capability and context. The technology is improving quickly, but financial services remains a high-trust environment with unique constraints around regulation, legacy systems, emotional customer interactions, skills gaps and fear among staff.

It cites a Microsoft report released last year, titled 2025: The Year the Frontier Firm is Born, which suggested that 82% of leaders believe this is a pivotal year to rethink operations, 81% expect agents to play a moderate or extensive role soon, 24% have deployed AI organisation-wide and only 12% remain in pilot mode.

For the sector, O’Keane argues, the challenge is not whether to adopt AI, but how to do so in a way that preserves trust, accountability and customer experience.

It adds that financial services organisations must contend with legacy processes, a high regulatory burden, emotionally charged customer interactions, skills shortages, vulnerability considerations and a gap between board-level enthusiasm and frontline caution.

As the ebook puts it: ‘These aren’t blockers. They are the reason a structured, human-centric approach is essential.’

What AI cannot do

A key thread running through the work is the idea that AI has clear limits, particularly in a sector where trust matters so much.

O’Keane says organisations’ values, aspirations and moral compass cannot be delegated to a machine. AI cannot replicate the abilities that underpin sound financial judgement, such as interpreting ambiguity, applying lived experience, setting organisational intent or finding creative solutions to nuanced customer problems.

For the sector, O’Keane states that ‘this balance is essential — too few humans, and decisions lose nuance; too few agents, and organisations cannot scale sustainably.’

That does not mean AI has no place in management or operations, but that human leaders ‘safeguard ethics, empathy, compliance, and contextual judgement.’

A framework for taking financial services to the frontier

At the centre of O’Keane’s thinking is Telefónica Tech’s Prism Framework, which he describes as ‘strategic assessment methodology designed to enable organisations understand their current position in the AI maturity journey and identify high-impact next steps.’

Rather than assuming every part of an organisation will move at the same pace, Prism treats the enterprise as a collection of components that mature at different rates. The idea is to allow leaders to zoom out, identify the most important opportunities and prioritise the places where AI can make the greatest difference.

The framework is presented as a way to break down the organisation into tiles or cubes within a prism, with strategic factors influencing how quickly AI can scale across different areas.

O’Keane claims that this matters because AI readiness is shaped by a range of forces, including culture, data maturity, risk appetite, regulatory context, brand promise and market pressures. These are fluid, it argues, and shift with leadership priorities, regulation, customer expectations and an organisation’s own transformation history.

To see a visualisation of the framework, be sure to check out the ebook (it’s free): The Human Side of Finance: Taking Financial Services to the Frontier.

The broader argument in the ebook is that of a ‘Frontier Firm.’

As the ebook states: ‘A Frontier Firm is characterised by high AI maturity, organisation-wide deployment, and a belief in the transformative power of AI agents. These firms operate with hybrid teams, i.e. humans and agents working side by side.The imperative to pursue Frontier Principles stems from the need to scale faster, work smarter, and create sustainable value in a landscape where AI is rapidly democratising access to expertise.

‘A Frontier Firm fuses human judgement with AI agents to scale faster, work smarter and unlock new value. In financial services, this is already happening: agents triage cases, assistants embed in work flows, and multi-step processes gain autonomous support.

Crucially, the Frontier does not replace humans — it elevates them. AI cannot replicate aspiration, judgement, or creativity. is not a place where humans are displaced. It is a place where they are elevated.’

Join the conversation

Hear more from Stephen O’Keane at our upcoming event, the FS Technology Summit, on 30th April in Edinburgh.

The conference will explore the transformative impact of technology on the modern Financial Services enterprise, examining how it can be applied to drive practical business improvement and frontline customer impact.

The conference is free* to attend (a criteria applies).

We expect the event to reach capacity, so book now to avoid missing out.

Secure your place here: FS Technology Summit | Scotland’s Biggest Financial Services Event