Your board approves an investment in artificial intelligence (AI). The budget is available, a technology partner has been selected, and the business case promises faster decisions and greater efficiency. Who, inside your organisation, will make it work?
That question was central to my presentation at the African Insurance Organisation (AIO) African Reinsurance Forum in Mauritius last October. My perspective comes from working in the African insurance and reinsurance talent market, where the availability and development of specialist people directly affect what businesses can deliver.
For an AI investment to succeed, someone must define the underwriting problem clearly. Someone must recognise when the available data leaves an important gap, or when a recommendation makes little commercial sense. And someone must remain accountable for the decision. Those responsibilities need to be considered while the investment is being planned.
The capability behind the technology
Boston Consulting Group (BCG) estimates that around 70% of AI’s potential value comes from change management, adoption and the redesign of talent and operating models. Its research is a useful reminder that the work required extends well beyond installing a system.
For leaders, this means asking whether the people expected to use AI have the knowledge, authority and time to do so effectively. An experienced underwriter may understand the risk exceptionally well but need support to assess how a model reaches its recommendations. A data specialist may understand the model but need help recognising where a treaty wording or local market practice changes the interpretation.
Bringing those people together is a management responsibility. Leaving them to address these gaps informally risks placing implementation in the hands of a few already overstretched individuals.
Designing for African markets
For African insurers and reinsurers, I would start with the capabilities required in the markets they serve. A talent plan needs to account for differences in data availability, language, distribution and the way risks are understood locally.
Consider a model supporting risk selection in a market where loss records are incomplete. The organisation needs people able to judge what the missing information means, what further evidence to seek and when to escalate a decision. Technical confidence alone cannot answer those questions.
Recruitment briefs should reflect that reality. Before advertising for an AI specialist, leaders should establish which business decisions need to improve and which capabilities are missing. They may need an insurance professional with stronger data fluency, a technical specialist with sector experience, or a leader who can help both groups work together. Each of these calls for a different approach to hiring and development.
Protecting the next generation of expertise
There is also a longer-term issue. Routine work has traditionally given junior insurance professionals repeated opportunities to learn. Reviewing files and discussing decisions with experienced colleagues helps them recognise patterns and build judgement.
BCG’s research highlights the need to redesign that learning as AI takes on more routine execution. My concern for employers is practical. If entry-level work changes, the route to becoming an experienced professional must change with it.
Leaders should make room for supervised case reviews and discussions about why an apparently reasonable recommendation might be wrong. Junior colleagues need opportunities to explain their reasoning, receive feedback and gradually take responsibility. A shorter processing queue is valuable, but organisations also need people capable of handling the difficult cases that remain.
A practical approach to closing the gap
In Mauritius, I used four routes to frame the response: Build, Buy, Borrow and Bridge. They provide a starting point for deciding how to secure the capabilities a particular AI initiative needs.
Build the capability of the people already in the business. Training should relate to the decisions they make each day, including how to question outputs, recognise limitations and escalate concerns. Managers need to give people time to practise those skills.
Buy capability selectively. Recruit externally where a critical gap cannot be closed internally within the required timeframe. Be precise about the work the person will own and who will support them, rather than expecting one appointment to solve the entire transformation challenge.
Borrow expertise through partnerships, specialist advisers, universities or technology providers. External support can help an organisation move forward while its own capability develops. The relationship should include a clear plan for sharing knowledge and retaining enough internal understanding to oversee the work.
Bridge the gaps between technical teams, insurance professionals and leadership. Assign responsibility for translating the business problem into technical requirements and for agreeing how decisions will be reviewed. People need to know who can challenge an output and who has authority to stop or change a process.
Putting people into the investment decision
A useful first step is to take one proposed AI initiative and map the people required to deliver it. Identify the capability already available, the gaps that could delay progress and the person responsible for closing each one. Give that plan the same attention as the technology budget.
The question I would leave with insurance and reinsurance leaders is straightforward. If your AI investment were approved tomorrow, could you name the people who would make it work, and explain what they need to succeed?
COMMENTS