Advisers balance AI adoption with human oversight

By Milan Rojan

Financial advisers have shown strong interest in using agentic AI for administrative tasks but have remained cautious about allowing the technology to make decisions that affect client money, according to a latest research report.

The research, based on a survey of 178 advisers, has highlighted a clear preference for using AI in high-volume, rules-based activities across the financial advice journey. Eight in ten respondents have been comfortable with AI collating information for annual reviews and suitability packs, while 77% have supported its use for onboarding and letters of authority.

A further 76% of advisers have been comfortable with AI supporting know-your-customer and anti-money laundering checks, while 75% have supported its use for fees and charges reconciliation.

However, support has fallen as AI has moved closer to decisions involving client assets. For pension transfers, 43% of respondents have been comfortable with AI handling tasks, while 29% have remained neutral and 29% uncomfortable. For CIP switching and rebalancing, 53% have expressed comfort, compared with 22% neutral and 25% uncomfortable.

The research has also highlighted a knowledge gap around agentic AI. Nearly one-third of advisers have been unable to describe what the technology does, while 29% have not identified its ability to break tasks into steps, plan workflows and complete processes automatically.

After being shown a description of agentic AI, however, 62% of advisers have said they would be comfortable with the technology being embedded within their existing platform, while a further 18% have remained neutral.

Rob DeDominicis, CEO of GBST, has said advisers have established a clear boundary for AI adoption. “They are comfortable with agentic AI taking on the high-volume administrative tasks, like reconciliation and collation of data,” he has said.

The findings have also indicated that advisers have preferred AI to operate within existing platforms and controls rather than as separate tools. GBST has said this approach has supported oversight and security while allowing firms to automate operational processes.

The research has reflected the wider debate around governance as financial institutions have expanded agentic AI into advice and wealth management workflows.