TL;DR
Get home office essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Capgemini’s World Cloud Report for Financial Services 2027 says multi-agent AI is moving into banking and insurance, but only 10% of surveyed firms have achieved enterprise-wide orchestration. The study identifies technology debt, governance and weak measurement of returns as barriers to scaling.
Financial institutions are moving multi-agent AI into banking and insurance workflows, but only 10% have achieved enterprise-wide orchestration, according to Capgemini’s World Cloud Report for Financial Services 2027. The findings point to a gap between experimentation and large-scale deployment, as firms contend with technology debt, governance needs and difficulty showing returns.
The report is based on a survey of 1,100 financial services executives across 14 markets. It says 59% of banks and insurers are piloting or deploying multi-agent AI, while only 10% have orchestrated the technology across the enterprise. The figures describe different stages of adoption: having a pilot or deployment does not mean a firm has connected AI agents across business-wide processes.
Capgemini reports that eight in 10 executives identify reducing technology debt as a priority for scaling agentic AI. Nearly half of firms cannot consistently demonstrate a return on their initiatives, and 51% lack a clear framework for deciding whether AI investments deliver enough value to justify ongoing spending. The report does not provide a financial measure of returns across the surveyed firms.
Firms are also putting more attention on control and cost. 90% use or plan to implement intelligent model routing, which directs tasks to models based on the intended balance of accuracy and cost. The survey says 52% identify hyperscalers as preferred partners for governing multi-agent systems, with regulatory certifications, security and identity frameworks, and auditability among cloud-provider selection considerations.
Why AI Adoption Has Not Scaled
The gap between 59% piloting or deploying multi-agent AI and 10% achieving enterprise-wide orchestration matters because firms can adopt agents for individual tasks without coordinating them reliably across operations. Wider use in banking and insurance could affect activities such as account servicing, lending and underwriting, but the report suggests that firms have not yet broadly established the systems and controls needed to manage those deployments across an organization.
The return-measurement findings also carry business implications. If nearly half of firms cannot consistently show returns and 51% lack a clear assessment framework, decision-makers may struggle to compare projects or justify continued spending. Capgemini says the funding focus is expected to shift toward business outcomes: the number of executives naming outcomes as their primary AI funding criterion is projected to rise 141% by 2028, while the number prioritising efficiency is projected to fall 42%. These are report projections, not observed results.
For customers, the study points to a possible expansion of AI agents into customer-facing responsibilities. Three quarters of respondents expect agents to become a key part of customer engagement within three years. That expectation is not evidence that such services have already been widely deployed, and the report does not specify how individual firms will handle customer consent, errors or human review.
Products Worth Considering
FLEXISPOT 36-Inch Standing Desk Converter, Black
FLEXISPOT 60-Inch Executive Standing Desk with Drawers and CPU Cabinet
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
As an affiliate, we earn on qualifying purchases.
From Individual Agents to Coordinated Systems
Capgemini’s report describes a change in the kind of AI deployment financial institutions are pursuing. Its global head of cloud for financial services, Ravi Khokhar, said firms had moved in 12 months from experimenting with individual agents on isolated tasks toward deploying systems that coordinate actions across business lines. That is his characterization of the sector’s direction; the survey figures also show that enterprise-wide orchestration remains limited.
The report identifies a small group, 5% of respondents, as having the capabilities to operationalise agentic AI at enterprise scale. More than half of that group orchestrate multi-agent systems across operations, technology and customer experience, compared with 8% of other firms. Capgemini also says 83% of these leaders link AI investment directly to business results. The report does not give further detail here on how the leader group was defined beyond its operational capabilities.
Cloud infrastructure is part of the proposed scaling approach, while sourcing models may combine internal development, packaged products and external partners. Capgemini expects this hybrid sourcing approach to rise from 36% of firms today to more than two-thirds within three years. The projection reflects executives’ reported expectations, not a confirmed future outcome.
Limits of the Survey Findings
The report summary does not provide the survey’s full methodology, the precise wording of every question, or market-by-market results. It is therefore unclear how adoption levels differ across countries, types of financial institutions or individual business functions. The headline figures are based on executives’ survey responses and should not be read as an independently audited count of deployed systems.
Several figures concern expectations rather than completed changes. This includes the projected shift in AI funding priorities by 2028, the anticipated rise of hybrid sourcing and respondents’ expectations for customer-facing agents. The source material does not specify how those projections were calculated or give a detailed timetable for individual use cases. It also does not report measured customer outcomes, cost savings or investment returns across the sector.
Where Financial Firms Go From Here
Capgemini’s report points to continued work on reducing technology debt, establishing governance and security controls, and linking AI spending to measurable business outcomes. It also expects firms to rely more on hybrid sourcing and to use intelligent model routing to manage model choice and costs. These are directions described or forecast in the study, not commitments from every institution surveyed.
The next meaningful test will be whether financial institutions can move beyond pilots and limited deployments to coordinate agents across business processes while maintaining oversight and showing results. The report offers three-year expectations for broader customer engagement and hybrid sourcing, but it does not identify specific institutions, deployment dates or public milestones that would establish whether those forecasts are being met.
Key Questions
What did Capgemini’s report find about multi-agent AI in financial services?
Capgemini says 59% of surveyed banks and insurers are piloting or deploying multi-agent AI, but only 10% have enterprise-wide orchestration.
What is preventing broader deployment?
The report points to technology debt, governance and security requirements, and difficulty demonstrating returns. It says eight in 10 executives view reducing technology debt as a priority, while 51% lack a clear framework for assessing AI investment value.
What does enterprise-wide orchestration mean in this report?
It refers to coordinating multi-agent systems at an enterprise scale, rather than using individual agents for isolated tasks. Capgemini’s summary does not provide a formal technical definition or a detailed threshold for classifying a firm as enterprise-wide.
Are banks already using AI agents for customer service?
The report says three quarters of respondents expect AI agents to become a key part of customer engagement within three years. That is an expectation, not confirmation that the use is already widespread.
How reliable are the report’s forecasts?
The survey records executives’ views and projections, including expected changes through 2028. Those projections are not guaranteed outcomes; the report summary does not provide enough methodological detail to independently evaluate how the forecasts were produced.
Source: finextra.com
