Mauritius in search of the right strategy

Local and international news have recently focused heavily on the idea that the ever-expanding application of AI in our lives could one day mean the end of humanity. For all the negative narrative around artificial intelligence, its promises remain enticing. For a rapidly changing economy like Mauritius, faced with a difficult international context, a contracting workforce and in permanent search of productivity gains, AI, if it is harnessed correctly, can be a tool impossible to ignore.

The use of artificial intelligence (AI) in the Mauritian economy is a source of much debate. While major world powers such as the United States and China have chosen to integrate AI into many pillars of their economies despite the risks often associated with it, it remains to be seen whether Mauritius will embrace AI to boost economic performance at a time when past certainties are quickly being eroded. 

For Julien Guillot‑Sestier, Co-founder of Eqxia (Applied Intelligence), the answer is an unequivocal yes. Most people, he explains, assume that a company adopting AI becomes a smaller company; meaning that it employs less people. Not so, he says, referring to the PwC 2026 AI Jobs Barometer which, having analysed more than one billion job advertisements across 27 countries, found the exact opposite to be true: the companies most exposed to AI increased their headcount by 52% over the period studied, while the least exposed grew by only 36%.

 

“The urgency is in upskilling and reskilling.”

 

Productivity figures point in the same direction: the industries most exposed to AI recorded a growth of 34% between 2018 and 2025, compared with 24% for the least exposed industries. Indeed, workers who report having AI skills benefit from an average wage premium of 62%, up from 57% a year earlier.

The incentive to integrate AI is therefore clearly there.

Pravesh Gaonjur, COO of Tylers, shares the same opinion. “Mauritius should view AI as a strategic opportunity, not merely as a technological trend. With labour shortages and skills gaps affecting several sectors, AI can help bridge part of that gap by automating repetitive tasks, improving service delivery and enabling better decision-making,” he argues. 

“If adopted with the right governance, infrastructure and investment in skills, AI can help Mauritius modernise faster, improve productivity and reinforce its ambition to become a reference economy in the region.”

Upskilling and reskilling

Building a more innovative economy and improving productivity through AI is an imperative that the government has clearly identified in the 2026-2027 Budget. Its importance is amplified in a challenging local and international economic context.

This is the view of Tylers’ COO, who argues that “AI, if utilised properly, can be extremely valuable in a difficult economic context. Indeed, it can help both government and businesses do more with limited resources by reducing manual processes, improving analysis and accelerating execution.” 

“For Mauritius, this can translate into higher productivity, stronger GDP growth and better management of public spending. However, the impact will depend on execution: AI must be linked to clear national priorities, measurable productivity gains and strong governance.”

“There is no day on which someone presses a button and AI arrives. It has been inside our phones and our navigation apps for years, improving quietly.”

For Julien Guillot-Sestier, identifying AI as a pillar to build on in the Budget was the right call, but we need to get the sequence right. “I want to say that plainly before adding anything else: the urgency is in upskilling and reskilling, and I would have liked to see that named as the priority.”

His rationale is that AI absorbs tasks within a job long before it affects the job itself. “An administrative role does not disappear. Six of its twelve tasks quietly stop requiring a person. You see this very clearly in business process outsourcing and in financial services, which happen to be two pillars of our economy. The work that remains is the part where human judgement is worth paying for, and moving people towards that part is the whole exercise,” he argues. 

To enable that to happen, people will need to be trained. For Pravesh Gaonjur, “the sooner the government, the private sector and the public service treat that as a shared challenge, the better prepared this country will be.”

Planning, training and a framework

According to Julien Guillot‑Sestier, successful AI implementation rests on three key pillars: data, process and people. 

“Every organisation that has been operating for ten years holds an archive nobody has ever really read,” he explains. “Sales exports, production logs, maintenance records, supplier invoices, years of spreadsheets that were built for one meeting and never opened again. A sugar estate, a textile factory, a hotel group and a small logistics firm all have that same archive, and the same reports that nobody has time to read properly.”

Until recently, interrogating that data meant hiring a consultant for three weeks and paying for a study that answered one question. With AI, the cost of asking a question of your own data has fallen by a factor he says he “would not have believed possible in 2022.”

“The good news is that it starts small. A sector-specific AI application is a project with a budget and a timeline. Reading your own numbers properly can start this week, with files you already own,” says the Co-founder of Eqxia.

There are, however, limits to what AI can achieve in that regard. “These systems work on data that exists and is reachable. Where a company never recorded something, no model will invent it. Where the data sits in four incompatible systems with no common reference, the first piece of work is unglamorous and has nothing to do with AI,” explains Julien Guillot‑Sestier.

Understanding what AI can and cannot do prevents a business under pressure from wasting money it does not have.

This reality, he argues, is particularly relevant for Mauritius, where many organisations already possess years of operational, financial and administrative data that often remain underutilised. Before investing in sophisticated AI solutions, businesses and public institutions must therefore ensure that their data is accessible, reliable and organised, while also preparing their workforce and internal processes for the transition.

“The country investing the hardest in artificial intelligence, and moving the fastest, is also the one drawing a legal line around what employers may do with it.”

Implementing AI across both the private and public sectors must be done gradually and in a phased manner. This is the view of Pravesh Gaonjur, who believes that Mauritius should begin with the most accessible applications of the technology. “In the first phase, Mauritius should prioritise software-based AI because it is generally less capital-intensive, easier to deploy and more mature than hardware-heavy automation,” he says.

According to him, sectors such as ICT, financial services, insurance, professional services and customer support can adopt AI rapidly to improve quality, reduce turnaround times and lower operating costs. 

For other sectors however, like agro-industry, the path forward can be less straightforward.

“Agro-industry can also benefit, but it requires a different transition path. Current farmers and operators would need training to understand and use AI effectively, while a new generation of tech-savvy farmers could explore models such as AI-assisted vertical farming, IoT-enabled irrigation and controlled-environment agriculture,” explains Pravesh Gaonjur.

Successful AI adoption, he believes, will not depend solely on technology itself, but also on careful planning, workforce training and a policy framework capable of supporting innovation over the long term. Feasibility studies, for example, will be essential to assess investment costs, local demand and competition from imports. 

“If Mauritius wants such initiatives to succeed, public policy must also create the right conditions so that local innovation is not undermined before it reaches maturity,” he insists. 

Bureaucracy reimagined 

Bureaucratic inefficiencies are among the areas identified by the government where the use of artificial intelligence could provide significant support to meet the expectations of citizens, who increasingly expect public services to be digital, simple and fast, instead of having to queue physically, complete paper forms and resubmit documents already held by government agencies.

For Pravesh Gaonjur, “identity verification, licensing, health services, banking onboarding and property transfers can all become more efficient if supported by secure digital platforms, responsible data sharing and technologies such as blockchain where appropriate.”

As such, “the benefits go beyond convenience: better digital public services can reduce traffic, improve productivity, lower administrative pressure and rebuild trust in the public sector. A smoother experience for citizens ultimately contributes to a more efficient and more confident economy.”

Although he shares the same view, Julien Guillot‑Sestier stresses the importance of getting the process right, as “digitising a bad process gives you a fast bad process. AI makes that mistake easier to commit, because it can automate steps that ought to be deleted. In public administration, where procedures accumulate for decades and nobody is ever asked to remove one, simplifying first matters more than anywhere else.”

If that is done right, the potential benefits are substantial, as highlighted by Raj Pentiah, the Minister of Public Service and Administrative Reforms, who recently said that if the public service responded properly to demand, Mauritius’ GDP could rise by 1.5%.

Where Julien Guillot‑Sestier is immediately optimistic is the document layer, because it requires no process reform at all. “Official correspondence in this country is often written in a register a large part of the population struggles with. Rewriting it in plain language now costs close to nothing, and it would change how citizens experience the administration faster than any new portal,” he explains.

AI in transport

Another sector where AI is already transforming things is public transport, although the Co-founder of Eqxia argues that most of us stopped noticing years ago. 

“Imagine Google Maps and Waze disappearing tomorrow morning. No rerouting around a traffic jam on the M1, no warning about an accident ahead, no arrival time you can plan around… That arrival time is itself an AI result: DeepMind rebuilt the prediction engine behind Google Maps and improved its accuracy by up to 50% in some cities. Nobody announced it,” explains Julien Guillot-Sestier.

The time gain effect was modest, but real. 

“That is exactly the point. There is no day on which someone presses a button and AI arrives. It has been inside our phones and our navigation apps for years, improving quietly,” adds our interlocutor. 

The really interesting work, he believes, starts one level up, as “analysing vehicle flows across a whole island tells you which junction creates a queue three kilometres away, or which bus route serves a demand that no longer exists.”

Not a replacement

AI integration in the transport industry will not come without pushback, as evidenced by the wave of anger set in motion by its use in establishing the roster of CNT employees. AI, says Pravesh Gaonjur, should not be presented as a replacement for human beings, as “it can replace or augment specific tasks, but it cannot replicate the full range of human judgement, accountability, empathy, creativity, and ethical responsibility.” 

According to the COO of Tylers, for now, the real priority is to redesign work so that people use AI to become more effective, rather than allowing AI to displace people without a plan. At the same time, we must remain realistic about long-term risks.

Julien Guillot-Sestier shares the same view that decision makers must “train your people, write your governance policy, and decide deliberately what you automate and what you keep.” 

Indeed, legal factors will also have to be factored in. Here, Julien Guillot-Sestier points to a case that was resolved in April 2026, when the Hangzhou Intermediate People’s Court ruled that a company could not dismiss an employee on the sole ground that AI had taken over their role.

“The case involved a quality assurance worker who had been demoted with a 40% pay cut when a system took over his work of checking model outputs, then dismissed when he refused reassignment. The court held that companies cannot unilaterally lay off staff or cut salaries because of technological progress, building on an earlier ruling from December,” he explains. 

The fact that the story comes from China is significant, he adds, in that “the country investing the hardest in artificial intelligence, and moving the fastest, is also the one drawing a legal line around what employers may do with it. Those two things are not in contradiction. They are the same policy.”

We should not demonise AI, but as Pravesh Gaonjur says, we must not stop questioning it either. “Every major technology creates supporters and sceptics, and that debate is healthy when it leads to better safeguards,” he notes.

Ultimately, AI is a tool: its impact depends on how it is designed, governed, and used. The priority should therefore be responsible adoption, with clear governance frameworks, security controls, accountability mechanisms, and safeguards against misuse.