The financial services sector has entered a decisive phase of its technological transformation. According to Capgemini’s World Cloud Report for Financial Services 2027, based on a large-scale survey of 1,100 executives in 14 international markets, what began as one-off automation projects is rapidly evolving into ecosystems of interconnected artificial intelligence agents capable of coordinating workflows, making decisions, and executing autonomous actions across the business. However, the report warns that the distance between ambition and real execution is deep: the proliferation of fragmented initiatives, legacy operating models, and siloed governance keep the vast majority of institutions trapped in pilot phases.
The Leap From Isolated Pilots To Multi-Agent Orchestration
The recent trajectory shows a dizzying but uneven pace of adoption. Whereas in 2025, 80% of institutions began designing or testing individual agents for specific efficiency tasks, 59% now say they are testing or deploying multi-agent systems in production within limited functions. However, only 10% of firms have achieved enterprise-wide multi-agent orchestration capable of sustaining fully autonomous business execution.
This transition goes far beyond marginal productivity gains. Agent ecosystems are beginning to reshape high-volume operational flows in banking and insurance, such as customer service, policy underwriting, claims management, and compliance operations. At the same time, user interaction is moving toward an intermediated model: 76% of senior executives expect artificial intelligence agents to become a key component of customer relationships within three years. Despite this momentum, Capgemini’s assessment through its maturity index, which measures dimensions such as deployment, trust, cloud authority, security, and operational discipline, finds that only an exclusive 5% of financial sector institutions currently have comprehensive enterprise maturity.
Three Strategic Priorities To Lead In The Agentic Era
The study, produced by the Capgemini Research Institute and presented by Kartik Ramakrishnan, chief executive officer of the firm’s Financial Services Strategic Business Unit, stresses that market leaders will not be defined by how many bots they deploy, but by how they orchestrate their autonomous execution, govern it at scale, and turn it into tangible results. To close the operational gap, the report identifies three key priorities:
It is vitally important to move beyond isolated use cases and integrate agents that span operations, the technology infrastructure, and customer journeys. In this regard, 44% of the executives surveyed already prioritize building “agentic meshes” to link scattered pilots with corporate execution. Likewise, firms are recalibrating their sourcing strategy: building in-house what generates strategic differentiation, buying standard solutions for generic needs, and partnering in ecosystems where shared data provides differential value.
The cloud is shifting from a simple provider of computing capacity to becoming the execution authority for autonomous systems, acting as the data spine, control plane, and governance fabric. When selecting cloud providers, regulatory requirements, data residency, and compliance certifications lead executives’ priorities (50%), followed by security and identity management (47%) and auditability or traceability at scale (39%). Moreover, sovereignty goes beyond the mere location of data: it covers direct control over who authorizes and executes agents’ decisions, a critical aspect given that 62% of firms acknowledge that their identity and access management infrastructure lacks the resilience required to operate continuously in autonomous environments.
As agentic workflows expand, pressure mounts to justify investments with quantifiable business returns. The report reveals a severe lack of financial discipline: 48% of companies cannot consistently demonstrate the return on investment of their agentic artificial intelligence initiatives, while 51% lack mature valuation filters (value gates) to authorize continued funding of projects. In response, the most advanced organizations are directly linking artificial intelligence spending and token consumption to concrete operational goals, establishing strict economic governance and shared accountability in the profit and loss statement.
