APAC finance firms eye $3m agentic AI savings

Only 6% said they were very familiar with the technology in compliance work.

Nearly half of Asia-Pacific financial institutions said agentic artificial intelligence (AI) could save their organisation between $2m and $3m a year in compliance spending.

Many see potential for agentic AI to cut compliance costs, but many firms are still unfamiliar with the technology and expect to keep human oversight in place, according to a Risk.net and Fenergo survey.

However, familiarity with the technology remains uneven. Only 6% of respondents said they were very familiar with agentic AI technologies in compliance, whilst 12% were moderately familiar. 

Another 53% said they were somewhat familiar, and 29% said they were not familiar with the technology.

The report said firms are still testing how agentic AI could be used in compliance work, including pulling information together and preparing first-cut assessments.

Concerns remain around explainability, privacy, data quality and the risk of relying too heavily on automation. 

The report said these issues are likely to keep human review in place even as firms expand their use of AI.

When asked how they expected to develop agentic AI systems, 55% of respondents said they were most likely to use a hybrid model combining off-the-shelf products with customised in-house platforms.

Another 24% said they would develop systems in-house, whilst 21% preferred vendor-led, out-of-the-box solutions.

The report also highlighted concerns over the “black box” nature of some AI systems, particularly where firms cannot clearly explain how automated decisions were reached.

The Risk.net/Fenergo survey was conducted between September and November 2025 and received 110 responses from risk, financial crime and compliance specialists at banks and asset managers, mainly in Singapore, Malaysia, Australia and the UK. A series of in-depth interviews was also conducted in November 2025.