South Africa’s insurance sector is entering a critical inflexion point, with Deloitte’s view being that that AI ambition across the industry will stall unless insurers accelerate cloud modernisation at the same time as their AI strategies are being implemented.
The report shows that although many insurers are piloting generative and agentic AI tools, most are still operating in fragmented, legacy environments that cannot support the scale, speed, and risk controls the technology requires.
Andrew Warren, insurance sector leader at Deloitte South Africa, says the next 18 months will determine which insurers can compete in an AI-driven market and which might be left behind.
“The industry has reached a point where experimentation is no longer enough. Insurers are starting to feel the limits of old architectures. Without deliberate cloud modernisation, AI will remain stuck in pilots, and insurers will struggle to meet emerging expectations for model oversight, ethical use, and operational resilience.”
A linked future
Cloud and AI cannot be treated as separate strategies. Cloud maturity determines how fast an insurer can deploy AI safely and at scale, while AI’s growing value proposition strengthens the case for faster cloud adoption.
When insurers pursue these strategies independently, they create structural bottlenecks that slow progress, increase risk and limit the returns on investment.
Warren says the competitive implications are significant.
“Insurers that modernise first will open new markets faster, reach underserved customers, and deploy AI with the confidence that regulators expect. Those that delay will find themselves constrained by technical debt, compliance pressure, and an inability to scale innovation beyond prototypes.”
Do not get left behind
The risk for insurers with slow cloud adoption is simple. AI decisions become harder to govern, audit, and secure, which puts both customers and the business at risk.
The report shows how some of the most significant gains will come from the everyday tasks that slow insurers down.
Agentic AI can handle historically person-intensive tasks, such as developing policy documents. With the right Agentic AI setup, those steps can shrink to minutes to produce a workable draft.
With the right “human-in-the-loop” activities to ensuring a correct final document, this speeds up the product development cycle. The same applies to customer service.
Automated agents can handle simple questions quickly and consistently, giving human teams the space to focus on the situations that need experience and judgment.
Claims processing is another area where the impact could be significant. AI can review documents, images, or videos, and flag the information needed to finalise a claim far faster than current systems allow.
That can translate into quicker decisions and a more predictable customer experience. AI can also help insurers untangle legacy systems by analysing legacy code and extracting the rules that keep the business running.
Critical expectations
Supervisors are sharpening expectations around technology risk, and insurers that cannot demonstrate strong control over AI will face the toughest scrutiny.
For insurers, aligning cloud, data governance, model oversight, and AI strategy is becoming a regulatory and competitive necessity. Regulators are also becoming far more interested in how insurers use technology.
Insurers with older systems or scattered cloud setups will feel this pressure first, as it is harder to demonstrate that everything is under control when the technology is spread across different locations.
Meanwhile, consumers will experience the difference in practical ways. Modern cloud platforms enable claims to be turned around faster and for service teams to provide more consistent answers.
Advisors can also draw on better information when guiding clients. But if insurers rush ahead without the correct checks, mistakes creep in. A policy could be misinterpreted, a claim could be misjudged, or advice could miss something important.
The report is clear that the safest route is to introduce AI gradually, keep people in the loop and build trust in the technology over time.
“AI is not the barrier anymore. Cloud is. And unless insurers fix that gap, the value they expect from AI will not materialise,” says Warren.
The report predicts that early signs of measurable value will begin to emerge next year.
The insurers that act now to modernise their cloud estates and design for optionality will be the ones positioned to integrate new AI breakthroughs, respond to regulatory expectations, and transform their businesses.
Warren says the choice is now strategic.
“Insurers that continue to delay cloud modernisation will not have the foundations they need to scale AI safely or competitively. The sector stands at a crossroads.
“Those who modernise first will move ahead. Those that wait will find themselves anchored to legacy systems that cannot support the next decade of innovation.”