Banks are becoming increasingly efficient at processing transactions, assessing applications, monitoring risks and managing customer interactions, but an important part of their operational capability may be quietly disappearing as more of these activities are automated. Experienced banking employees have traditionally carried knowledge that extends well beyond documented procedures, including an understanding of why certain controls exist, how unusual transactions should be handled, where operational weaknesses are likely to emerge and when a seemingly routine situation requires additional scrutiny. Much of this expertise develops through years of dealing with exceptions, resolving problems and understanding the consequences of decisions across different banking functions. As software increasingly performs routine tasks and artificial intelligence begins supporting more complex activities, employees are becoming less directly involved in the processes through which this knowledge was originally acquired. The immediate benefits of automation are clear, but the longer-term implications deserve closer attention. Financial institutions may eventually find themselves operating highly sophisticated systems with fewer people who fully understand the business processes behind them, creating a new challenge for operational resilience, workforce development and the ability to maintain effective human oversight.
Banking Knowledge Has Never Been Entirely Written Down
Banking operations depend on a combination of formal procedures, regulatory requirements, technology systems and professional judgement. While institutions maintain extensive documentation covering everything from payment processing and credit assessment to customer verification and compliance monitoring, the practical knowledge required to operate these processes has historically developed through experience.
An experienced payments specialist, for example, may recognise that a particular transaction pattern requires closer examination even when it does not trigger a formal exception. A credit operations employee may understand why certain supporting documents are important in circumstances that differ from standard lending applications, while a compliance professional may recognise inconsistencies in customer information that would be difficult to capture through a simple checklist.
These judgements are not necessarily the result of employees disregarding established procedures. They often reflect a deeper understanding of how those procedures apply in different circumstances, particularly when information is incomplete, situations are unusual or several operational requirements must be considered together.
This type of expertise is commonly described as tacit knowledge, meaning knowledge developed through experience that is difficult to express fully through written instructions. It can include recognising early warning signs, understanding informal dependencies between departments, knowing which questions to ask and identifying when an apparently successful process has produced an unexpected outcome.
For many years, banks transferred this knowledge through practical experience, supervision, mentoring and exposure to increasingly complex responsibilities. Junior employees learned routine processes before progressing to exceptions and more difficult decisions, developing an understanding of how banking operations functioned beyond the individual tasks assigned to them.
Automation is changing that learning environment. When technology handles the routine work that previously provided the foundation for professional development, employees may enter supervisory or decision-making roles without acquiring the same depth of operational experience.
Automation Is Removing the Work Through Which Expertise Was Developed
The efficiency gains associated with banking automation are substantial because many operational processes involve repetitive activities that can be performed more quickly and consistently by software. Document classification, transaction matching, customer information checks, payment validation and routine account servicing are examples of activities where automation can reduce processing times and improve productivity.
However, these activities have historically served another purpose: they helped employees understand how banking processes work.
A junior operations employee who manually reviewed payment instructions would gradually become familiar with transaction formats, settlement requirements, common errors and the relationships between different processing systems. Someone working through customer onboarding applications would develop an understanding of documentation requirements, customer circumstances and the reasons certain applications needed additional verification.
As these tasks become automated, employees may increasingly encounter only the cases that technology cannot resolve. While this can make their work more intellectually demanding, it also changes the way they develop expertise because they are being asked to handle unusual situations without necessarily having extensive experience of the routine processes underlying them.
The problem becomes particularly important when organisations assume that employees can supervise automated decisions simply because they understand the relevant policies. Effective supervision often requires practical familiarity with the process being automated, including an ability to recognise when an output is technically plausible but operationally incorrect.
A bank may therefore reduce the amount of manual work required to operate a process while unintentionally weakening the experience base needed to oversee that same process.
The Most Experienced Employees May Be the Last to Understand Certain Processes
Another challenge arises as experienced banking professionals retire, change roles or leave their organisations. Many have worked through several generations of banking technology, giving them an understanding of how current systems evolved and why particular processes were designed in certain ways.
These employees may know that a specific reconciliation procedure exists because of a historical processing issue, or that an apparently unnecessary approval requirement addresses a risk that is no longer obvious from the current workflow. They may also understand how different departments interact when standard processes cannot be followed.
When these individuals leave, institutions do not necessarily lose access to formal documentation, but they can lose the context needed to interpret it effectively.
This becomes more complicated when legacy banking processes have been automated or integrated into newer platforms. Business rules originally developed by experienced employees may be embedded in software configurations, scripts or decision engines, with limited documentation explaining the reasoning behind them. Over time, the organisation may retain the technology that performs a function while losing the people who understand why it performs that function in a particular way.
The risk is not confined to older systems. Newer cloud-based applications and AI-enabled workflows can create similar dependencies when implementation teams, consultants or specialist employees possess knowledge that is not transferred effectively to the wider organisation.
For banks, this creates a form of institutional knowledge risk that may not appear in conventional operational dashboards. A process can continue functioning reliably for years while the number of employees capable of explaining, challenging or redesigning it gradually declines.
Human Oversight Becomes Less Effective Without Human Understanding
The expansion of AI into banking makes the preservation of operational knowledge particularly important because financial institutions are increasingly expected to maintain meaningful human oversight of automated systems.
In principle, human oversight provides a safeguard against inappropriate decisions, unexpected outcomes and circumstances that automated processes cannot adequately interpret. In practice, however, oversight is only effective when the people responsible possess sufficient knowledge, authority and information to challenge the technology.
Consider an AI-supported credit assessment process that produces a recommendation based on customer information, financial records and established lending criteria. An employee reviewing the recommendation may be able to confirm that the required fields are complete and that the output falls within the bank’s policy parameters. Yet identifying a more subtle problem, such as an unusual income pattern, inconsistent supporting evidence or a risk not adequately represented in the underlying data, may require experience developed through reviewing real lending cases.
The same principle applies to financial crime monitoring, customer complaints, payment investigations and regulatory reporting. Employees who rarely perform the underlying activities may become increasingly dependent on the automated system’s interpretation of the information, making it harder to identify errors that fall outside predefined controls.
There is also a behavioural dimension. When automated systems consistently produce reliable results, employees can become accustomed to accepting their recommendations, particularly when workloads are high and the technology is presented as more accurate or efficient than manual processing. Over time, supervision may become a procedural confirmation rather than an independent assessment.
This is not an argument against automation or AI. It highlights the importance of designing human oversight around demonstrable expertise rather than simply retaining a person somewhere in the approval process.
Banks need to ensure that employees responsible for challenging automated decisions continue to develop the knowledge necessary to do so, even when technology performs most of the underlying work.
The Next Skills Gap May Be in Banking Operations, Not Technology
Discussions about the future banking workforce frequently concentrate on shortages of AI engineers, data scientists, cybersecurity specialists and cloud architects. These capabilities are undoubtedly important, but the growing emphasis on technical skills can obscure another requirement: employees who understand banking operations deeply enough to translate business needs into effective technology and control arrangements.
As banks modernise, they need professionals who can explain how a payment should move through different stages of processing, what constitutes an acceptable lending decision, why a compliance control exists and how operational exceptions should be resolved. This expertise is necessary not only for running existing services but also for designing, testing and improving the automated systems that increasingly perform them.
Technology teams can build sophisticated applications, but they depend on business specialists to define requirements, validate outcomes and identify operational consequences that may not be apparent from technical specifications. If the bank’s operational knowledge becomes concentrated among a shrinking group of experienced employees, technology development can become more difficult even as software engineering capabilities improve.
The problem is particularly relevant to AI implementation because models and agents require clearly defined business rules, reliable evaluation criteria and an understanding of situations in which automated decisions should be restricted or escalated. Without experienced operational specialists, banks may struggle to determine whether an AI system is genuinely improving a process or simply performing its existing steps more quickly.
This suggests that the future banking skills challenge is not a choice between technical expertise and traditional banking knowledge. Institutions need employees capable of combining both, particularly in roles involving process design, operational risk, technology governance and automated decision-making.
The most valuable professionals may increasingly be those who understand how banking processes work, how technology changes them and where human judgement remains essential.
Banks Need to Preserve Knowledge Before Automating It Away
One response is to treat knowledge preservation as part of the automation lifecycle rather than an activity undertaken only when experienced employees are preparing to leave.
Before automating an operational process, banks can examine how employees currently perform the work, including the decisions they make, the exceptions they encounter and the informal knowledge they use to resolve unusual situations. This provides an opportunity to capture expertise that may not be reflected in formal procedures or technology requirements.
The objective should extend beyond recording individual steps. Banks need to understand why decisions are made, what risks different controls address and how employees determine when a case requires additional attention.
For example, documenting a payment investigation process should involve more than identifying the systems an employee accesses and the sequence of checks performed. It should also capture how experienced investigators distinguish between different types of processing failures, which inconsistencies suggest a more serious problem and how they determine whether a transaction can be safely corrected.
Similar approaches can be applied to lending, customer onboarding, financial crime investigations and regulatory reporting, particularly where specialist judgement plays an important role.
Knowledge capture also needs to remain current. Banking regulations, products and technologies evolve, meaning that historical practices cannot automatically be treated as authoritative. Institutions should distinguish between valuable operational experience, outdated workarounds and procedures that require redesign.
AI may help with this process by organising operational documentation, analysing historical cases and making institutional knowledge easier to retrieve. However, the accuracy and relevance of captured information still need to be validated by qualified employees, especially where it influences financial decisions or compliance obligations.
Simulation Could Become an Important Part of Banking Workforce Development
If automation reduces employees’ exposure to routine operational work, banks may need to create alternative ways for them to develop practical experience.
Simulation offers one possibility. Rather than relying entirely on live transactions and customer cases, institutions could develop controlled environments in which employees practise handling unusual situations, investigate operational discrepancies and evaluate decisions made by automated systems.
A payments team, for example, could work through simulated scenarios involving incomplete transaction information, conflicting settlement records or unusual processing patterns. Credit operations employees could review complex lending applications containing inconsistent information, while financial crime specialists could investigate cases designed to test their ability to recognise risks that automated monitoring systems have not fully identified.
These exercises would serve a different purpose from conventional compliance training because the emphasis would be on developing judgement rather than simply demonstrating knowledge of procedures. Employees would need to explain their reasoning, identify alternative interpretations and understand the consequences of different decisions.
Simulation could also help banks identify where operational knowledge is concentrated. If only a small number of experienced employees can resolve certain scenarios, the institution gains evidence that additional training, documentation or succession planning may be necessary.
Technology providers have an opportunity to support this approach through training environments, realistic test data, scenario-generation tools and platforms that allow employees to examine how automated systems respond to unusual circumstances.
As banking operations become more automated, the ability to practise situations that employees rarely encounter in normal work may become an increasingly important component of professional development.
Technology Vendors Have a Role in Preserving Banking Expertise
Banking software providers are often evaluated according to their ability to reduce manual processing, improve accuracy and accelerate business workflows. These measures remain important, but they do not fully capture how a technology solution affects the knowledge and capabilities of the employees responsible for operating it.
A system that automates a process without explaining its decisions may deliver short-term efficiency while making it harder for employees to understand how outcomes are produced. Conversely, software that provides clear decision histories, accessible business rules and meaningful explanations of exceptions can help preserve operational understanding even as manual activity declines.
This creates opportunities for vendors to differentiate their products through transparency, training capabilities and support for human judgement.
For AI providers, the challenge is particularly significant because users need to understand not only what a system recommends but also the information, rules and limitations relevant to that recommendation. Explainability alone may not be sufficient if employees lack the underlying expertise to interpret the explanation, making user training and operational knowledge development important parts of implementation.
Banks may also increasingly expect technology vendors to support knowledge transfer during major transformation projects. This could involve documenting business logic, training internal specialists, providing access to configurable rules and ensuring that important operational knowledge does not remain exclusively with external implementation teams.
The commercial opportunity is therefore broader than automating existing processes. Providers that help financial institutions retain and develop operational expertise may offer greater long-term value than solutions focused solely on reducing human involvement.
Operational Resilience Depends on Knowledge That Technology Cannot Replace
The disappearance of human knowledge is unlikely to produce an immediate or easily measurable crisis. Banking processes may continue operating successfully, automation rates may increase and productivity indicators may improve even as practical expertise becomes concentrated among fewer employees.
The consequences may emerge gradually, particularly when institutions introduce new products, modify business processes, respond to unusual customer circumstances or need to challenge the behaviour of automated systems. In these situations, the ability to understand how banking operations function can be as important as the technology used to execute them.
This creates a different perspective on operational resilience. Banks must preserve more than infrastructure, applications and formal procedures. They also need to maintain the organisational capability to interpret events, make informed judgements and adapt when established processes no longer provide sufficient guidance.
The answer is not to retain unnecessary manual work or prevent automation from improving banking efficiency. It is to recognise that operational expertise is an institutional asset that requires deliberate investment, particularly when the activities through which employees traditionally developed that expertise are disappearing.
Banks that incorporate knowledge preservation into technology transformation, provide realistic opportunities for employees to develop judgement and maintain strong connections between operational specialists and technology teams may be better positioned to manage increasingly automated environments.
Ultimately, the success of banking automation should not be measured only by how much human work can be eliminated. It should also consider whether the institution retains the knowledge required to understand, govern and improve the processes that technology now performs.
The long-term risk is not that banks will become too automated, but that they may become so dependent on automation that fewer people understand the banking activities taking place beneath it.
What this means for the industry
- Operational knowledge should be treated as an institutional asset. Banks need to identify and preserve expertise that exists beyond formal procedures, particularly where it is concentrated among experienced employees.
- Automation strategies must account for workforce development. Removing routine tasks can improve efficiency but may also reduce opportunities for employees to acquire the experience needed to handle complex cases.
- Human oversight requires demonstrable expertise. Employees responsible for reviewing automated decisions need sufficient operational understanding to challenge outcomes rather than simply approve system recommendations.
- Knowledge capture should form part of technology transformation. Documenting business rules, decision reasoning and exception-handling practices can reduce the loss of institutional expertise during automation projects.
- Simulation and practical training may become more important. Controlled operational scenarios can help employees develop judgement even when routine banking activities are largely automated.
- Banking technology vendors have an opportunity to support knowledge retention. Transparent decision processes, accessible business rules, training environments and effective knowledge transfer could become important product differentiators.
- Operational resilience requires a balance between automation and understanding. Banks need to preserve the human capability to interpret, challenge and improve automated processes as their operating models evolve.
