Banks and insurance companies are more likely to work with suppliers to develop artificial intelligence (AI) agents than develop them in-house.
Two-thirds of more than 1,000 respondents to Capgemini Research Institute’s World cloud report in financial services 2026 said they work with external partners for AI expertise and cloud-based platforms.
For example, health insurance provider Vitality has recently deepened its ties with Google through the launch of an AI platform designed to help millions of people make better-informed decisions about their health and lifestyle.
Meanwhile, Japan’s largest bank, Mitsubishi UFJ Financial Group (MUFG), aims to transform into an AI-native company by using agentic AI, changing how it handles data, and inking key partnerships with OpenAI and Sakana AI.
Only 33% said they are actively developing proprietary AI agents. Almost half (49%) are combining in-house development with supplier offerings, and just 15% are buying AI agents off the shelf, said the report.
Huge growth to come
The Capgemini Research Institute estimates that AI agents could bring up to $450bn in economic value by 2028, and with only 10% of banks and insurers deploying AI agents at scale, there is a huge opportunity.
“AI agent adoption is poised for rapid growth as 80% of financial services firms are in the [idea] or pilot stage of deployment,” said the report. “However, a sizeable opportunity remains to be unlocked.”
Ravi Khokhar, global head of cloud for financial services at Capgemini, said: “Our data reveals widespread industry optimism that the agentic era will open doors to new markets, signalling a new phase of transformation is upon us. To realise this potential, financial institutions must take a long-term view as humans work alongside agents. This means separating substance from hype.”
Commenting on the report, Jörgen Olofsson, chief information officer at Euroclear Sweden, said: “AI agents provide financial institutions with significant opportunities to automate repetitive tasks, improve customer and business support functions, and boost operational scalability.”
To date, the survey found that 75% of banks use cloud-native AI agents for customer services, 64% for fraud detection, 64% for loan processing and 59% for customer onboarding.
In the insurance sector, 70% use AI agents for customer service, 68% for underwriting, 65% for claims processing and 58% for customer onboarding.
Jesse Antosiewicz, senior director of technology at Liberty Mutual Insurance, who took part in the survey, said: “AI is already delivering measurable improvements in customer experience. From faster processing to more intuitive interactions, the technology is reshaping how services are delivered and how customers engage with the brand.”
Another survey respondent, William Tong, senior vice-president of the business technology payments service at Capital One, said: “Agentic AI is gaining trust in customer-facing roles as firms strengthen policies and governance. Opportunities lie in accelerating long processes such as onboarding customers, merchants and partners, shrinking cycle times to transform efficiency and elevate the overall customer experience.”
AI babysitters
Despite AI’s widespread use, the survey found that about half of banks and insurers are creating AI supervisor roles as 92% of executives admit to a lack of relevant skills among business leaders. AI agent supervisor and coordinator roles are being created by 48% of those interviewed.
As well as creating AI agent supervisory roles, 46% of banks and insurers are reskilling employees and reallocating them to different divisions according to skill sets.
Lloyd Scholz, chief technology officer at insurer Markel, said: “Maintaining a human-in-the-loop remains crucial, especially in regulated settings. Trust, explainability and governance are vital components in deploying AI agents, ensuring that automation supports rather than replaces human judgment.”