Banks Want Agentic AI But Can't Automate Basic Tasks
Ninety-one percent of banking innovation leaders believe agentic AI will reshape financial services. Yet fewer than one in ten banks have fully automated foundational operations like processing direct debits or posting daily interest.
A survey of 150 UK banking leaders managing balance sheets between £0.5 billion and £100 billion reveals the core problem: ambition has outpaced infrastructure. Only 31% are actively deploying AI in operational or decision-making processes today, despite widespread conviction that the technology matters.
The gap exists because agentic AI-systems that reason, decide, and execute multi-step workflows autonomously-requires clean, real-time data flowing through modern systems. Banks still run on legacy infrastructure that chokes on both.
What's Blocking Progress
Three barriers stand in the way. Seventy-seven percent of leaders cite legacy systems restricting data availability. Another 77% point to poor data quality. Seventy-one percent struggle to access real-time data.
The manual operations problem cuts deeper. Between 37% and 42% of institutions still rely on manual workarounds for core tasks:
- Standing orders and scheduled payments: only 10% fully automated
- Daily interest accrual: only 11% fully automated
- Account maturity instructions: only 13% fully automated
- Scheduled interest rate changes: only 13% fully automated
Sixty-one percent of respondents describe these basic tasks as "very" or "extremely painful" to execute. Among organizations with minimal automation, 85% report high pain levels. For those with full automation, the figure drops to zero.
Regulators Won't Accept Black Boxes
Deploying autonomous agents creates regulatory risk that banks cannot ignore. The UK's Financial Conduct Authority and the US Consumer Financial Protection Bureau demand strict accountability, data lineage, and models that prevent discriminatory outcomes.
Seventy-nine percent of innovation leaders recognize this: without high-quality, explainable data, AI could worsen financial exclusion rather than improve it.
Yet only 12% feel confident they could explain and justify AI decisions to regulators today. If an agent denies a loan or freezes an account, the core banking system must surface an immutable audit trail showing which real-time data points drove that decision. Legacy systems cannot reliably do this.
Fix Core Systems First
The business case for agentic AI is real-sophisticated wealth advisors, automated credit underwriting, and autonomous decision-making in front-office functions all promise operational gains.
But layering complex AI on top of manual, fragmented infrastructure compounds risk. Banks cannot automate the future using the workflows of the past.
The path forward requires core modernization. Migrating from rigid, siloed legacy systems toward cloud-native, data-driven engines eliminates manual friction and creates the clean data flows that agentic AI needs to function safely and compliantly.
Only after achieving 100% automation on basic tasks can banks add agentic AI and see real operational returns. Learn more about AI agents and automation, or explore how operations teams can prepare for AI-driven transformation.
Your membership also unlocks:
