Sector
AI leadership for financial services
Last updated
Justas Butkus is an AI expert and fractional AI officer based in Vilnius, Lithuania, working with mid-market companies and scale-ups across the UK, EU and US at two to four days a month. He builds and operates the production AI systems he advises on, rather than only advising on them.
Short answer
Financial services has the two conditions that make AI worth doing: high-volume repeating processes and expensive qualified people doing them. It also has the condition that makes it hard – every automated decision must be explainable to someone who can fine you for it.
Where does AI actually pay for itself in financial services?
The processes worth automating in any sector share three properties: they repeat constantly, they consume qualified people, and nobody enjoys them. In this sector that usually means:
- Client onboarding and KYC review – document collection, checking, chasing and re-checking.
- Inbound query triage – routing, classifying and answering the same questions at volume.
- Reconciliation and exception handling – where the exceptions, not the matches, consume the day.
- Arrears and collections contact – high volume, tightly regulated, poorly staffed at scale.
- Internal knowledge retrieval – staff hunting through policy documents for answers that exist.
None of these are exciting. That is the point – the value comes from volume, not sophistication, and the exciting use case is usually the one that fails.
What makes this sector different
Two things change the design here rather than merely adding paperwork.
- Explainability is not optional. A decision affecting a customer must be reconstructable later, which constrains architecture from day one rather than being retrofitted.
- Automated decision-making has its own rules under data protection law, separate from the AI Act, and they predate the current AI conversation entirely.
- Procurement diligence is the real gate. Most firms meet AI governance requirements through a customer or insurer questionnaire long before a regulator asks.
The practical consequence: governance designed in at the start costs a fraction of governance retrofitted after legal stops a launch.
How an engagement here typically runs
- Inventory what is already runningIncluding the unapproved tools. In regulated environments this is usually the most uncomfortable and most valuable step.
- Map the repeating processes against costFrequency multiplied by fully loaded cost per occurrence. The ranking usually surprises people.
- Establish governance proportionate to the riskWho approves what, what happens when a system is wrong, and how a decision is reconstructed months later.
- Ship the unglamorous one firstProve the pipeline end to end before attempting the ambitious thing.
Frequently asked questions
Where does AI pay for itself in financial services?
In high-volume repeating processes: onboarding and KYC review, query triage, reconciliation exceptions, collections contact and internal knowledge retrieval. The value comes from frequency rather than sophistication.
Does the EU AI Act classify financial services AI as high risk?
Some of it. Creditworthiness assessment and risk pricing for life and health insurance are named in Annex III. Much operational automation is not high risk, which is why classification is an early piece of work rather than an assumption.
What usually blocks these projects?
Rarely the technology. Usually inconsistent data definitions across systems, or the absence of anyone willing to accept the risk of putting a system in front of customers.
If this describes your operation
The useful first conversation is about one specific process rather than about AI in general.