Affordability files, arrears and forbearance packs, statutory notices and complaint chronologies: that is where AI agents earn their place in UK consumer lending operations. Lending decisions, forbearance outcomes, vulnerability judgements and final complaint responses stay with people. The FCA's consumer credit rules, the Consumer Duty, SM&CR and UK GDPR set the frame.
Where does a UK consumer lending operations team spend its time?
Affordability evidence. FCA rules require a proportionate creditworthiness and affordability assessment, and lenders now expect to show their working: income verified from open banking data or payslips, committed expenditure from bureau and transaction data, and a judgement about what the borrower can sustain. Operations assembles that file and checks it is complete.
Arrears and forbearance. Customers who fall behind must be treated with forbearance and due consideration, and those with characteristics of vulnerability need particular care. Teams handle payment arrangements, breathing space notifications, income and expenditure assessments and referrals to debt advice. Every step has to be recorded because the FCA and the ombudsman will ask.
Letters and complaints. Statutory notices, annual statements, arrears notices and default notices must go out on time with the right content, and late or wrong notices can stop a lender enforcing. Complaints, especially about affordability and historic lending, generate large case files that someone must read in full.
Where can AI agents help?
- Affordability file assembly (depth 2). The agent reads transaction data and payslips, categorises income and committed spending under the lender's policy, flags gambling transactions or payday borrowing where policy treats them as signals, and drafts the affordability summary. An underwriter decides.
- Income and expenditure preparation (depth 2). For customers in arrears, the agent pre-fills the income and expenditure assessment from available data so the conversation with the customer starts from facts. A collections colleague completes it with the customer.
- Statutory letter production (depth 3). The agent generates arrears and default notices and annual statements from the loan system on approved templates, checks timing against the rules the lender has documented, and dispatches routine notices within limits. Any notice for an account flagged as vulnerable or in a breathing space is held.
- Complaint case packs (depth 2). The agent assembles the lending decision, the data used, the contact history and the product terms into a chronology, and drafts a summary of what the file shows. A complaint handler writes the response.
Breathing space is where agent limits matter. When a customer enters a statutory debt respite, contact and enforcement must stop. An agent that sends a routine arrears letter in that period causes a breach. The control is a hard bound: no outbound collections letter on an account with an active breathing space flag, and the verdict filed. The approach to proving those bounds before going live is in deploying AI agents: train, shadow, certify, widen.
What stays with people, and why?
Lending decisions stay with underwriters. The FCA expects lenders to make reasonable, evidenced affordability judgements, and the Consumer Duty asks whether the outcome is good for the customer. A person deciding on a complete file is how you answer both.
Forbearance decisions stay human: what arrangement to offer, whether to freeze interest, whether to write off. So do vulnerability judgements. An agent can flag that a customer mentioned a bereavement or a diagnosis; a trained person decides what adjustment follows.
Final responses to complaints stay with a handler, with redress decisions signed off by someone with authority. Under SM&CR, a named senior manager is accountable for these processes, including how agents are used within them.
Which rules apply in the United Kingdom?
The FCA regulates consumer credit firms and sets the rules on creditworthiness, arrears, forbearance, vulnerable customers and complaints. The Consumer Duty requires firms to deliver good outcomes for retail customers across products and services, price and value, consumer understanding and consumer support. Our piece on FCA Consumer Duty and AI agents in lending goes through each outcome. The PRA supervises banks prudentially and its SS1/23 sets model risk management principles that reach agents used in credit processes. The Senior Managers and Certification Regime makes named individuals accountable.
UK consumer credit legislation sets content and timing for statutory notices, and errors can affect enforceability. Lenders should map each notice type to the controls that check it.
UK GDPR and the Data Protection Act 2018, overseen by the ICO, govern borrower data, including the protections on decisions based solely on automated processing with significant effects and the duty to carry out a data protection impact assessment for high-risk processing. Transaction data reveals sensitive information such as health spending, and the agent's access should be limited to what the assessment needs.
The UK takes a principles-based, regulator-led approach to AI rather than a single AI law, so these existing rules are what apply.
This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.
How do you start in the first 90 days?
Weeks 1 to 4: start with complaint case packs. They are read-and-draft work, the handler checks everything, and they save hours on every file. Agree the case pack structure with your complaints lead first.
Weeks 5 to 8: shadow statutory letter production. The agent generates each notice, the existing process sends the real one, and you compare content and timing. Test the breathing space and vulnerability holds deliberately.
Weeks 9 to 13: certify letters at depth 3 for standard accounts, and start shadowing affordability file assembly. Brief the accountable senior manager on the bounds. For a comparison with US rules on credit denial, see our brief on operations agents in US consumer lending and ECOA and AI agents in lending.
Questions people ask
Can an agent categorise gambling transactions in affordability checks?
Yes, if your policy defines how gambling spend is treated and the agent applies that definition. It drafts the categorisation and the underwriter weighs it. Be careful that the agent does not infer more than the data shows.
Who is accountable under SM&CR when an agent sends a wrong notice?
The senior manager responsible for the process, as with any other system or team. That is why bounds, certification and the filed record matter: they are how that person shows they took reasonable steps.
Can the agent detect vulnerability from call notes?
It can flag phrases and patterns for review. It should not label a customer as vulnerable on its own, because that label changes how they are treated and needs a trained person's judgement.