Dive Brief:
- Nearly half of the financial organizations creating multiagent systems within core banking and insurance workflows are unable to consistently demonstrate returns on their investments, according to a report from the Capgemini Research Institute. The findings are based on a survey of 1,100 C-suite executives at financial firms.
- While agentic AI systems are being piloted or deployed by almost 60% of the financial institutions surveyed, only 1 in 10 has successfully orchestrated agents to run autonomously across their environments, according to the report.
- Successful multiagent systems require an established cloud foundation, which serves as the execution layer governing agentic data access, decision-making and actions, according to Ravi Khokhar, global head of cloud for financial services at Capgemini. “As agentic AI scales, those with the strongest foundation will see the true value,” he told CIO Dive.
Dive Insight:
As financial institutions expand into orchestrated, multiagent workflows, CIOs need to establish the proper setting through cloud infrastructure and cost management systems.
Almost all of the financial institutions surveyed are using or planning to use automated model routing to help keep token costs in check, according to Capgemini. Model routing sends queries to the most cost-effective AI models that can deliver the appropriate quality output. Organizations are expecting that cloud providers will invest in building out ways to enable model selection, Khokhar said.
Bank of America, for example, implemented Orchestra, an AI routing layer that uses a small, classifier AI model to send queries to the most financially-effective model. Meanwhile, Travelers Insurance also relies on model routing to help manage costs. The insurance giant built its own large language model, TravelersLLM, to handle queries at a lower cost than frontier models.
“The cloud foundation is seamlessly able to offer that routing without losing the context and without losing the access to the right data,” Khokhar said. “That is what is unlocking what we call return on intelligence.”
But financial institutions are likely to continue running some AI workloads on premises, too. The survey found that a hybrid approach to cloud and agentic AI capabilities will become the dominant approach for more than two-thirds of financial organizations within three years, according to Capgemini.
“A hybrid strategy is a composable ecosystem where you use fit-for-purpose language models, fit-for-purpose compute environments, and fit-for-purpose cloud environments,” Khokhar said.
Along with disciplined cost management tactics, AI funding decisions are shifting away from operational efficiency gains and toward business outcomes, according to Capgemini. The firm projected that the number of executives emphasizing business outcomes as a justification for AI spend will jump 141% by 2028.
