Nigeria’s insurance market is preparing for a shift in the way companies calculate the capital they need to hold against underwriting risks. Business A.M.’s October 19 report describes the proposed risk-based capital framework under development at the National Insurance Commission, with Ernst & Young advising. The change aims to make the sector more resilient.
Capital protects an insurer’s ability to meet its obligations. Yet customers experience another kind of risk at the claims desk: an incorrect assessment, a confusing rejection, or a decision nobody can clearly explain. As companies add AI to operations, those failures can become faster and harder to unravel.
Nigeria should use the wider reform conversation to set an operational expectation. Every consequential AI-assisted recommendation that influences a claims decision should leave a record that a trained employee can inspect and a dissatisfied policyholder can challenge.
A human assessor could receive an AI summary of a motor policy, claim photographs, repair invoices, and previous correspondence. The summary might omit a document or misunderstand an exclusion. It could also flag a legitimate claim as suspicious because the claimant’s evidence follows an uncommon pattern. These are illustrative risks, not allegations against any named Nigerian insurer. What matters is whether the assessor has time and authority to verify the evidence before the result reaches the customer.
I would require three things from companies choosing to deploy these systems.
The first is a decision record. Keep the information relied on, the AI-generated recommendation, material corrections made by staff, and the person responsible for the final determination. That record should survive a model update or vendor change. Firms need a secure way to retain relevant evidence without unnecessarily exposing customer data.
The second is a genuine route to reconsideration. When a claim is disputed, a reviewer who did not simply approve the original recommendation should examine the underlying policy language and evidence. A consumer must be told in ordinary terms why a decision was made, what information can change it, and how to request a human reassessment. An appeal that reruns the same automated logic offers little protection.
The third is a measurable operating standard. Quarterly management reviews should track the share of AI-assisted decisions requiring correction before issue, the number of appeals, the proportion overturned after independent examination, average time to resolution, and patterns affecting particular products or groups of customers. These figures need not disclose individual policyholder details. Managers should use them to improve workflows and training, not merely to celebrate faster processing.
There is a commercial reason to take this seriously. Business A.M. reported KPMG’s October insurance research: globally, insurers are experimenting with AI, but very few respondents reported fully redesigned claims and policy servicing operations. That finding describes KPMG’s surveyed organisations, not a measured adoption rate for Nigeria. It still warns against treating an automation pilot as proof that customer-facing processes are ready for unattended operation.
Nigeria’s own regulator has made trust a priority. In October discussions with the Nigerian Insurers Association, the National Insurance Commission (NAICOM) highlighted policyholder confidence, operational efficiency, and risk management as priorities for market development. The Commissioner has also emphasised transparent claims processes and responsible technology in industry remarks.
These commitments deserve practical procedures at branch level. An employee needs permission to reject a model suggestion, a supervisor needs a clear basis for review, and a policyholder needs a way to contest the result. If staff fear that correcting AI decisions makes their productivity numbers look worse, those corrections will be hidden rather than used for learning.
NAICOM need not prescribe a single AI product or claims platform. It can ask insurers using AI-assisted decisions to demonstrate that each material recommendation remains reviewable, that appeal outcomes are recorded, and that boards can see error patterns alongside efficiency gains. The industry can adapt the details to different product lines and operating models.
Nigeria’s reform will strengthen financial resilience if the capital framework works as intended. A parallel test of reviewability and customer recourse would strengthen the relationship that gives insurance its purpose: the promise that valid claims receive fair treatment when policyholders need help.
- business a.m. commits to publishing a diversity of views, opinions and comments. It, therefore, welcomes your reaction to this and any of our articles via email: comment@businessamlive.com
PR shapes stakeholder expectations — without them realising it