Meta’s Muse and OpenAI’s Dots are among the first consumer-facing AI assistants built to handle everyday chores such as booking reservations, ordering groceries, and managing payments on a user’s behalf. Since its launch on September 8, Muse has been downloaded 9 million times and is reported to have 1.8 million daily active users according to AppMagic. Meta’s chief AI officer Alexandr Wang told the Financial Times that the company intends to bring these agents to billions of people worldwide, with plans to roll the service out beyond the United States and Canada.
Tackling Diverse Market Challenges
The rollout confronts a core challenge: the systems must work effectively in markets that differ sharply from affluent, English-speaking regions. Start-ups such as South Africa’s ManyHands, which offers the WhatsApp-based assistant Manny, are already testing the limits of that ambition. South Africa’s linguistic environment includes 11 official languages, and the real test lies in adapting to local communication styles, payment habits, and cultural expectations. Anton Moulder, ManyHands’ founder, explained to Rest of World that teaching an AI to handle these variations is far from straightforward.
Large language models, which power these assistants, often perform poorly in languages that receive less training data. English speakers therefore stand to benefit most, while users in the Global South risk being left behind. Payal Arora, a professor of inclusive AI cultures at Utrecht University, frames the issue as a matter of influence: “Linguistic inclusion is ultimately about power.” Without robust support for local tongues and cultural context, agents may impose Western assumptions on users, deepening economic and cultural gaps. Independent AI policy researcher Usha Raman warns that this could force non-Western users to adapt to the agents rather than the agents adapting to them.
Meta has acknowledged that expanding internationally will require integration of local languages and partnerships with regional services. Even if language hurdles are cleared, deeper obstacles persist. Many economies in the Global South depend on informal networks, cash-based transactions, and context-dependent communication—areas where AI assistants, trained primarily on Western data, may falter. For example, a grocery order in Lagos or Nairobi might involve negotiations, family preferences, or last-minute changes that a rigid system could overlook.
Infrastructure Strain Under Rising Traffic
Beyond language and culture, the sheer volume of assistant activity could strain already limited infrastructure. Tests by Cisco indicate that AI assistants generate up to 450% more traffic per task than a human performing the same action. Cloudflare reported a 1 700% increase in daily assistant requests over the past year. In regions with spotty connectivity—such as Niger, Burkina Faso, or eastern Congo—this surge could cause slowdowns, failed transactions, or outright outages. Kyle Spencer, executive director of the African IXP Association, notes that even areas with spare capacity, like East Africa, may encounter disruptions if undersea cables or local fiber networks are damaged.
Cloud-based assistants such as Muse and Dots can continue operating when a user’s connection drops, but the experience degrades noticeably. Approval requests or payment confirmations may time out, leaving tasks unfinished. Small offline vendors, in particular, would need to adapt to bot interactions, a shift that can be costly and technically demanding. Jane Munga, a fellow at the Carnegie Endowment for International Peace, cautions that this dynamic could deepen inequality. According to the International Telecommunication Union, 36% of Africans used the internet in 2025, compared with 92% of Europeans and 88% of North and South Americans. Urban users are 2.6 times more likely to be online than rural residents, meaning assistants could further concentrate digital access in cities.
Cost Barriers Threaten Adoption
Cost presents another barrier to widespread adoption. While Muse is offered for free, competitors such as Dots charge a $100 monthly subscription, a sum that exceeds the budget of most consumers in developing nations. Munga points out that even free assistants may entail indirect expenses, including data usage and transaction fees, which could exclude low-income users. The risk, she argues, is that AI assistants become tools reserved for the connected elite, widening the digital divide rather than narrowing it.
Support Costs, Data Inference, and Regulatory Gaps
Businesses may need to raise customer-support spending as agents generate extra purchases and returns. Human users can flag erroneous agent actions, prompting more refunds and stressing support teams. Agents improve when they have detailed knowledge of a user’s favorite restaurants, shops, and frequented locations. Such data helps the assistant form a picture of individual tastes, habits, and daily routines.
Regulatory Gaps Raise Accountability Concerns
Agents can infer attributes such as likely income, health status, creditworthiness, or behavior from processed data. When these systems are built and hosted abroad, local regulators may lack the ability to supervise them. This supervisory gap raises concerns about accountability and compliance with domestic policies. The lack of oversight can hinder enforcement of data-privacy standards.