Wealth managers need shared, current client records to shorten account opening because faster artificial intelligence tasks alone leave delays between teams unresolved, Databricks said in an analysis published Oct. 30.
Databricks, which sells a platform for data and AI applications, said improvements within individual systems can leave the broader process stalled. Its blog post, “Wealth Management Client Onboarding Needs More Than AI,” argues that firms must connect evidence, outstanding problems and decision histories across teams.
The company examined coordination problems in client onboarding, the process of collecting information and completing checks before opening and funding an account. It drew on industry studies published from 2024 through 2026, rather than presenting a new survey of its own.
Databricks cited Capgemini’s World Wealth Report 2026, which found that 60% of 144 wealth executives lacked a unified view of clients. The same study found that operational work consumed 41% of the time of 1,317 relationship managers across 16 markets. That measure covered wealth operations broadly, rather than onboarding alone.
A 2024 EY study of more than a dozen global asset managers found that more than half involved at least 10 teams in onboarding. Only 15% reported strong integration with customer relationship management systems. A separate 2024 study by KPMG and the Private Wealth Management Association covered 35 Hong Kong private wealth institutions and linked onboarding demands to rising costs and worsening client experience.
Databricks put private banking account opening in Singapore at about six weeks, with complex cases taking longer. The Monetary Authority of Singapore said in May that initiatives from a workgroup it co-led with industry were expected to bring the median below one month, the company said.
The authors also cited a 2025 EY survey of 100 decision-makers, split evenly between wealth and asset management firms. It found that 95% of firms were expanding generative AI across multiple uses, while 78% were exploring AI agents. Only 27% of executives reported business benefits from generative AI.
The studies covered different markets, groups and regulatory systems, so their findings require context, the authors said. They nevertheless identified a common problem: teams repeatedly searching for evidence, reconciling information and resolving outstanding issues as cases moved between systems.
The company proposed a shared case record that keeps earlier versions and links information to its source. It said this would let staff and supervised AI agents see completed checks, outdated evidence and unresolved issues without rebuilding the case history.
Databricks said a supervising agent could then assign tasks, use approved tools and bring in a human reviewer when judgment was needed. Core client and custody records could stay in their existing systems, while its platform handled evidence, checks and workflows, with record ownership and decision authority preserved.
