Beyond Cost Savings: Where AI Actually Changes Day-to-Day Banking

Quick Summary

Most AI research fixates on productivity and cost reduction. Deutsche Bank''s Tsvetanka Nankova argues the more interesting gains sit across the client value chain, from routing enquiries and clearing false positives in transaction monitoring to forecasting cash flows and generating new revenue.

Why is the cost-saving framing too narrow?

Nankova reads a great deal of consulting research on artificial intelligence, and notes that the benefits are almost always presented as improved productivity and lower cost. Her own view is that the opportunity runs across the whole value chain for clients, delivering a better experience and, importantly, new revenue flows. That reframing changes which projects get funded: a cost case tends to produce automation of existing tasks, while a revenue case produces new propositions.

What is AI already doing inside bank operations?

Operations teams handle a heavy load of enquiries and investigations, and Deutsche Bank already uses AI to make sure client enquiries are routed to exactly the right colleagues across the firm, so they are resolved professionally and quickly. The larger gain is in transaction monitoring, where the combination of AI and the structured data made available by ISO 20022 has produced a significant reduction in the number of alerts generated. The bank can clear false positives swiftly and concentrate on the alerts that carry real financial risk, which protects clients and, Nankova argues, protects the wider financial industry from undue risk.

Where does the revenue upside sit?

On the revenue-generating side, the applications are concrete:

• Cash flow forecasting, which allows the bank to schedule and execute payments on a client''s behalf.
• Liquidity management optimisation, which treasurers at both corporates and financial institutions are actively asking for as the world becomes more complex and fragmented.
• Client behaviour analysis, producing solutions tailored to individual client needs rather than a blanket approach across the whole portfolio.

That last point is the connective tissue: the same models that reduce noise in operations are what make genuinely differentiated service possible at portfolio scale.

FF NEWS TAKE:

The ISO 20022 detail deserves more attention than it gets. Years of migration pain are now showing up as a measurable drop in false positives, which is the clearest evidence yet that the data standard, not the model, was the bottleneck. Banks still treating ISO 20022 as a compliance project are sitting on the asset that makes the AI case work.