AI Has Changed – So Your Target Operating Model Must Change as Well

Introduction

We have moved past the “hype” phase of AI. The question is not whether to use AI, but how to embed it across the organisation. The challenge is that AI of 2026 – adaptive, generative, increasingly autonomous – looks nothing like the AI of a few years ago.

In banking and across multiple industries, AI is no longer an emerging capability on the periphery of organisations. It is already embedded across core activities such as credit decisioning, financial crime detection, customer service, treasury, risk management and software delivery. Yet many banks are still attempting to deploy AI within operating models designed for a pre-AI world. The outcome is predictable: fragmented use cases, unclear accountability, increased risk exposure, slower time-to-value and growing tension between innovation and control.

The technology has evolved, and your operating model must evolve with it.

Target Operating Model – a quick refresher

A Target Operating Model (TOM) defines how an organisation is structured, governed and run to deliver its strategy. A well-defined TOM typically addresses:

Processes: How work flows end-to-end.
People and skills: Roles, capabilities and capacity.
Technology and data: Systems, platforms and information assets.
Governance and controls: Decision rights, risk ownership and policies.
Organisation and sourcing: Structural design, third parties and partnerships.
Performance management: KPIs and value tracking.
Regulation and policy: Embedded compliance and reporting.

Traditional TOMs assume human‑led processes supported by deterministic technology. AI fundamentally disrupts this model.

Why AI breaks traditional operating models?

Traditional banking TOMs were built for “human-led, technology-assisted”. Modern AI flips this to “technology-led, human-governed”. That shift introduces challenges that legacy TOMs were never designed to accommodate:

Non-deterministic behaviour that complicates explainability, accountability and control.
Blurring of roles between humans and machines fundamentally changing ownership and ways of working.
Continuous learning and change requiring ongoing monitoring, governance and intervention.
New risk vectors including bias, hallucinations, data leakage, and regulatory scrutiny.

Trying to bolt AI onto an existing TOM typically results in tactical fixes, new committees, additional controls, or isolated centres of excellence, rather than a coherent, scalable model.

Designing a TOM for the era of AI

An AI-enabled TOM does not start with technology. It starts with reimagining how value is created when AI is a core participant in the operating model. The leading shifts we see:

Process: from automation to orchestration

Processes must coordinate human judgement, AI actions and clear escalation paths.

People and skills: new roles, not just new tools

Banks need AI Product Owners, model stewards/validators, Prompt Engineers/AI Trainers, and business leaders accountable for AI outcomes (not just deployments).

Technology and data: platforms over point solutions

Scalable AI platforms, strong data foundations, and integration with core banking systems replace isolated use cases.

Governance and risk: embedded, not bolted on

Risk management must be built into lifecycle management, change processes, monitoring and ownership aligned to regulatory expectations.

Organisation and sourcing: federated, but aligned

Banks will not centralise all AI capability – nor should they. The challenge is enabling federated delivery across business lines with consistent standards, tooling, and controls, plus enterprise-level oversight.

Why this matters now?

Regulators are tightening their expectations. Customers want more, faster. Margins remain under pressure. And AI is evolving at a pace most operating models simply cannot keep up with.

Banks that don’t evolve their target operating model risk ending up with:

Advanced AI capabilities they cannot deploy safely or at speed.
Rising operational and model risk.
Competitors who move and learn far more quickly.

Those that do adapt will unlock something far more valuable than efficiency: the ability to build institutional intelligence at scale.

How we help

We help UK banks and financial services firms design, evolve and implement AI-enabled operating models that bring together people, processes, governance, and technology into a single, integrated ecosystem. By clearly defining roles, responsibilities, decision rights, and workflows, we build an operating backbone that enables scalable, repeatable, and accountable AI deployment, while meeting regulatory and risk expectations.

What we do

TOM assessment: Identify friction between current structures and AI ambitions.
AI TOM design and implementation: Principles, target states, roles, workflows and controls that embed AI safely at scale.
Workforce transformation: Map current against needed skills; design training for reviewers, overseers and complex case holders.
Governance framework: Build agile guardrails that satisfy regulators (e.g. PRA/FCA) without stifling innovation.
Data and platform architecture: Design scalable AI platforms, solid data foundations and core-system integration.
Change and adoption: Support cultural change, ways of working, and value tracking (KPIs) over time.
Capability and maturity assessment: Benchmarking using our Operational Maturity Model (OMM) to sequence investments and ROI realisation.

What it achieves

Fluid structures that break down silos.
Governance that moves at the speed of code.
A workforce focused on oversight and judgement.
Faster time-to-value with clear accountability and auditable control.

The bottom line

AI is no longer a tool you buy and install; it is a capability you cultivate. It has already rewritten how value is created in financial services, and the real test is whether your operating model has kept pace.

A future-ready TOM doesn’t ask, “Where can we use AI?” It asks, “How should our bank operate when intelligence is embedded everywhere?” If your TOM cannot answer that, it is time to redesign it, not incrementally, but fundamentally.