At a contact centre, an agent takes calls as an AI tool “listens” with her. The tool suggests responses, flags possible quality issues and helps draft wrap-up notes. Training for this agent role has shifted: less time is spent on basic checks and more time on judgement calls and empathy. This prompts the question: do entry-level jobs disappear as the more routine tasks are automated by AI, or do they become a ladder to different roles? This question matters, especially in South Africa. Recent research by The Bridgespan Group, conducted in partnership with Harambee Youth Employment Accelerator, looked at the effects of AI on employment in the business process outsourcing (BPO) and tourism sectors of South Africa. We found that AI is mostly supporting workers, rather than replacing them. This could change, though, if the country doesn’t take action now that turns AI into an employment ally rather than an enemy. To do this South Africa would need to redefine roles, adapt skilling systems and evolve talent-matching.AI is reshaping, not replacingThe fear that AI will replace humans in the workplace is shared across the globe. Many industries are already experiencing disruption. In South Africa, where the broad youth unemployment rate for those aged 18-35 is 54.7% and 9.3-million young people in the same age group are not in employment, education or training – the anxiety is acute.Yet our research is cautiously optimistic. Employers and government officials in both BPO and tourism say AI is helpful, but that people remain essential. In practice, that means basic, repeatable tasks are increasingly handled through automation and self-service, while more complex customer experience work becomes a bigger share of what employees do.“Efficiency is not what customers remember,” notes Neville Cousins, digital, ITO, and PMO executive at Business Process Enabling South Africa (BPESA). “A human agent doesn’t just solve a query; they create reassurance, empathy, and trust. Customers know when they’re engaging with AI, and while it may be fast it often feels transactional.” In tourism, AI is similarly strengthening workers in the moments that matter most to customers, allowing them to combine booking, customer support and guest experiences.The findings led us to model two scenarios, defined by how AI is adopted, how demand will evolve, and how each industry might prepare for greater AI use. In the first scenario, strong public-private partnerships lead to proactive investments and co-ordinated preparation for AI, and strengthen South Africa’s competitiveness in both BPO and tourism. Demand rises, leading to more jobs in hybrid roles where people work alongside AI. Under this scenario we estimate employment in BPO could reach over 500,000 jobs by 2035 if organisations transition from isolated pilots to coordinated, large-scale AI adoption that blends human capabilities and automation. In tourism, there will be over 1.28-million jobs created by 2030, with adoption patterns and customer preferences changing where and how people work.The second scenario is characterised by limited preparation and reactive adaptation to AI’s impacts. AI adoption focuses on reducing labour costs without redesigning roles or building skills pipelines. Entry-level work shrinks, and South Africa will be less competitive than international locations that invest in AI. Under this scenario, BPO jobs created would be 250,000–290,000 jobs by 2035. Overall employment in tourism remains stable with under 1.19-million jobs created by 2030, as tourists continue to visit South Africa. However, the impact will be uneven across subsectors. AI is more likely to reshape roles where automation is easier, for example, travel bookings, while employment in roles such as transport remains steady.To be sure, AI is also creating space for new entrepreneurial roles, particularly in tourism. AI makes it easier for content creators, experience hosts and virtual tour curators to create and market their own offerings. But only if entrepreneurship is encouraged. “To enable equitable participation in self-employment,” says Harshita Shyam, AI/ML product manager at Expedia Group, “it is critical to simplify business registration for creators, reduce or defer fees until minimum revenue thresholds are met, and provide standardised digital pathways that enable micro-entrepreneurs to formalise quickly and access markets.”The aggregate figures in the two scenarios also hide important differences in employment opportunities. Women and young people are concentrated in the roles that typically serve as entry-level jobs: frontline service, administrative support and other positions structured around repeatable tasks. They are also more exposed to AI-enabled task automation.Even when overall employment remains stable, the composition of work can change in ways that narrow entry points for women and youth, especially if routine service lines shrink faster than new roles are made accessible. The risk, then, is a transition gap: if employers raise skill expectations faster than training systems adapt, entry-level workers can be locked out of the roles that remain.Three practical moves — and who can drive themThere is still a window of time to manifest the first scenario, which leads to more employment and a more competitive nation. Three practical moves are on the table:Build adaptive skilling systems: training providers, employers, funders, and government can shift trainings to emphasise digital literacy and practical exposure to the human capabilities that complement automation, such as empathy, complex communication, judgement and problem-solving. They can also show trainees what AI-enabled workflows look like, incentivise organisations to invest in entry-level talent, and integrate learning into entry-level work.Redefine roles for human-AI collaboration: employers and industry bodies can clarify hybrid roles and expectations, especially where people add value through empathy, escalation handling, complex communication and judgement, and create career pathways for workers whose tasks are shifting.Evolve talent matching and entrepreneurship pathways. Job-matching and gig platforms, placement organisations, funders and policymakers can improve the matching of employees into AI-enabled roles and provide stronger support for tourism entrepreneurship and gig-based opportunities, paired with safeguards and business support that reduces precarity for new entrants.AI adoption in South Africa’s BPO and tourism sectors is not yet widespread, which is precisely why this moment matters. With deliberate choices that strengthen people, lift service quality and protect entry pathways, particularly for women and youth, AI can become a productivity and inclusion story that simultaneously sustains South Africa’s competitiveness in the global services economy.• Nair is chief technology officer of Harambee Youth Employment Accelerator. Naidoo, Porteous and Mohapatra are with NPO The Bridgespan Group.