- within Criminal Law and Intellectual Property topic(s)
Why operating model and change management together decide whether AI investment in the Finance function pays back.
Global enterprise AI spend will pass $2 trillion in 2026, yet around 95% of AI pilots deliver no measurable P&L impact. The barriers to achieving a measurable impact are not just technological, but structural. For CFOs, the question is not whether to invest in AI but whether the business is structured to convert that investment into results.
Data integration, value capture, capital discipline, governance, and incentivisation are the levers that turn AI investment into measurable returns. The Finance function owns these levers, which puts the CFO in a strong position to drive measurable impact. The risk is treating AI as a procurement decision, when the binding constraints sit in the operating model and in the human layer around it.
The current state: pilots without payback
Most organisations are running AI pilots that fail to scale. The AlixPartners 2026 Disruption Index makes the European picture concrete: only 8% of European CFOs report enterprise-wide agentic AI deployment, while 47% are either still in pilot or making no use of agentic AI at all. Investment continues regardless: 52% of European CFOs are spending more on digital tools than a year ago. Spending is rising while conversion to operational reality is not.
Part of the reason is structural. Only 30% of European CFOs say their core systems are modern and fit for purpose, and 27% report teams running workarounds all the time. Globally, AI-leading organisations are nearly twice as likely to have an estate free of significant legacy issues. Each additional layer of unaddressed legacy processes and tech debt slows AI adoption further. Tools get bolted onto these workflows, function-led initiatives create disconnected islands with duplicate vendors and siloed use cases, and governance, risk, and incentives lag the technology. Value is rarely underwritten upfront, tracked monthly, or implementations stopped when it falls short.
Finance is well placed to break that pattern. The general ledger, planning systems and KPI hierarchies are already the cleanest source of truth in the enterprise. AI scales when built on that foundation, but only if the function redesigns itself first rather than layering AI over the existing shape.
What has to change: operating model and human layer, in parallel
The future-state Finance function does not look like a smaller version of today’s function. Functional silos collapse into value-stream pods organised around outcomes such as transaction integrity, insights and partnering, controls and risk, and agent operations. Routine work moves to an agent fabric that operates continuously rather than in cycles. Layers compress and new roles emerge inside the pods, including Agent Operators, Data Stewards, and Scenario Architects.
AI technology itself is increasingly commoditised; the harder problem is the human layer. Adoption, trust, capability-building and talent transition are where value is won or lost. The Disruption Index bears this out: 40% of European CFOs say the pace of change is making employees’ skills rapidly obsolete. The CFO’s objectives for an AI implementation span a spectrum, from aggressive cost-out at one end to employee-first reskilling at the other, with managed transitions in between. Alignment before implementation, and the quality of execution that follows, are more important than the specific objective chosen.
The risk of drift is not theoretical. The Disruption Index shows only 27% of European CFOs say they drive disruption rather than react to it, but 53% report feeling less anxious about disruption than they did a year ago. Leading the organisation through the holistic transition to an AI-based organisation is more important than the specific use cases legacy functions are testing. Without a deliberate position, the default outcome is inertia, which in AI translates into widening competitive exposure over time.
What CFOs should do now
- Lead with the operating model, not the technology. Redesign workflows without human-era constraints, then select tools that fit the requirements. It is a warning sign when sub-functions procure AI tools in silos.
- Discontinue legacy processes and supporting technologies. Stop performing redundant processes ‘just in case’, plan rapid switch off of legacy technologies, and resolve data quality at source through systems alignment.
- Deliberately position your change approach. Decide where you sit on speed of automation, transparency of communication, redeployment commitment, and governance alignment. Make it a choice, not a drift.
- Underwrite value upfront. Every use case needs a P&L target, a named owner, and a monthly measurement plan, with the discipline to stop initiatives that fall short.
- Name a single accountable owner for the human transition. Change management requires collaboration across HR, IT, and the PMO, but it must also have a single owner in each area.
Preparing for the future
The CFO’s role in AI to architect conversion rather than gatekeep spend, ensuring investment delivers measurable impact through accountability and governance. Real impact requires the operating model and the change layer to be redesigned in parallel. Within the larger organisation, the CFO can lead by example.
European CFOs already sense the gap. The Disruption Index shows 32% are pessimistic about AI’s impact, against around 18% across the broader executive group. That scepticism is rational. Understanding the true drivers of enterprise AI value, and acting on them, will separate the CFOs who realise the value from those who do not. Done well, the CFO can drive the proof cases for enterprise AI within the organisation, creating the foundation on which the rest of it scales.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
[View Source]
