South African businesses are adopting artificial intelligence at varying speeds, with many firms using only the basic AI features bundled into existing software while a smaller group are building purposeful in-house AI capability and a tiny vanguard explore advanced models, according to a sector-wide assessment by KPMG.
Three clusters of adoption
KPMG’s practical look across its client base groups organisations into three adoption buckets: 60% that use AI tools embedded in software but with mixed uptake; 35% that have established governance and are developing in-house AI use cases; and a leading 5% who are experimenting with advanced models and bespoke tools.
The largest cohort — the 60% — tends to run AI via packaged tools such as Copilot that come with office and productivity suites. Adoption in this group is inconsistent: some teams embrace the tools and learn quickly, while others barely use them. KPMG points out that governance questions, regulatory caution and uncertainty over a clear strategic direction often slow broader rollout.
From point solutions to purposeful programmes
Firms in the 35% band have moved from curiosity to commitment. They have assessed risks, set governance guardrails and started creating concrete use cases. Typical applications in this group focus on three core functions:
- Finance — automation and streamlining of repetitive processes;
- Human resources — workflows such as recruitment, onboarding and employee support;
- Sales — lead scoring, customer engagement and proposal generation.
KPMG notes technology, financial services and consumer/retail sectors — where firms are data-rich and already tech-enabled — are most likely to have active use cases across all three areas. In other industries uptake is often limited to one or two functions where the business case is clearest.
Vanguard firms use advanced models
The most advanced 5% are experimenting with models such as Anthropic’s Claude and Google’s Gemini, constructing bespoke tools and applying AI beyond back-office efficiencies to tasks like scenario analysis, simulation and modelling.
| Adoption group | Share | Typical characteristics |
|---|---|---|
| Embedded-tool users | 60% | Use bundled AI features; mixed uptake; governance not mature |
| Active builders | 35% | Defined governance; in-house agents; use cases in finance, HR, sales |
| Vanguard | 5% | Advanced models; bespoke tools; scenario analysis and simulation |
What this means for South African firms and households
For businesses the headline is straightforward: many companies are not yet extracting the full value from software they already pay for. That ‘latent capability’ is effectively deferred productivity — which matters because improved productivity is required to raise wages, support jobs and lift competitiveness in South Africa’s constrained growth environment.
Several practical constraints shape the pace of adoption:
- Governance and compliance concerns that delay rollout, especially in regulated sectors;
- Uneven digital and data maturity across organisations, which limits where AI can be applied effectively;
- Skills gaps in prompt engineering, model oversight and data management, which mean benefits concentrated where in-house technical capacity exists.
Where firms move from experimental projects to scaled deployment, the potential effects include lower unit costs, faster transaction processing and improved customer service. Those gains, in turn, affect household budgets through price, service quality and employment trajectories — though KPMG cautions that benefits will be uneven while adoption remains patchy.
Policy and boardroom implications
Boards and executives need a clearer ‘North Star’ for AI: a strategy that aligns risk appetite with measurable use cases and an investment plan for skills and governance. Regulators and industry bodies will also play a role in defining standards that allow innovation while protecting consumers and data privacy.
For South African businesses the KPMG assessment is a prompt to audit current software licences and evaluate whether untapped AI features can deliver short-term efficiency gains, while building the institutional capacity to move into the 35% or 5% categories over time.
WE NEWS does not provide financial advice; businesses and readers should consult their own advisers before making investment decisions related to AI strategy or procurement.
