What AI Voice Assistants Taught Me About Personalised Learning

My name is Jeffrey Mdala, and I am an AI Engineer and founder of Zambian Online Education Company (ZOEC), the company behind eskulu, an AI-powered learning platform built for the Zambian ECZ curriculum. Over the years, I have spent a lot of time thinking about one big question: how can technology adapt to people in a way that feels natural, useful, and genuinely supportive?

That question came to mind again while reflecting on something simple but powerful: using AI voice assistants like ChatGPT and Gemini for long-form conversation. What stood out to me was not just the novelty of talking to AI for an hour, but the deeper lesson behind it: people engage more when technology meets them where they are.

In Zambia and across Africa, that idea matters more than ever. We are building digital systems for real people with different personalities, learning styles, emotional needs, and social realities. If AI is going to make a meaningful impact here, it cannot be cold, rigid, or disconnected from everyday life. It has to feel responsive, adaptive, and human-centered.

Why AI Voice Conversations Feel Different

One thing I have observed is that AI in voice mode can hold someone in conversation for a surprisingly long time. In many cases, that happens because AI is designed to adapt to the user. It responds in a way that feels engaging, patient, and personalized. Instead of forcing the user into a fixed communication style, it adjusts.

That adaptability is one of the most important strengths of modern AI systems. When someone feels heard, they stay engaged longer. When responses feel natural, conversation flows more easily. And when the interaction is non-judgmental, people often become more comfortable expressing themselves.

As someone building in education, I find this extremely important. Traditional systems often assume that every learner should move at the same speed, respond in the same way, and understand concepts through the same format. But that is not how real people learn. Some learners need repetition. Some need encouragement. Some need voice. Some need text. Some need structure, while others respond better to exploration.

This is exactly why I built eskulu with the belief that learning technology in Zambia should not just digitize content. It should adapt to the learner.

What This Means for Education in Zambia

I started building education tools because I saw a real gap. Many learners in Zambia need affordable, accessible, and high-quality academic support, especially for ECZ subjects. That is why I built platforms like Zedpastpapers, which now serves over 200,000 users every month, and why I developed eskulu during the COVID-19 period.

When schools are under pressure, teacher-to-student ratios are high, and access to quality materials is uneven, AI can do more than automate tasks. It can become a support layer around the student. A voice-based AI assistant, for example, can help a learner ask follow-up questions without fear of embarrassment. It can explain the same concept multiple times. It can shift tone and pacing depending on the learner's needs.

In the African context, this matters because educational inequality is not just about whether content exists. It is also about whether learners feel able to engage with it consistently.

That is one reason eskulu has been built to offer:

• Free notes for Grades 6–12
• Past papers and marking schemes
• Quizzes and revision support
• An AI tutor experience designed around the Zambian curriculum

For me, the future is not simply putting textbooks online. The future is building systems that can respond intelligently to the learner in front of them.

AI Should Feel Helpful, Not Intimidating

One of the most exciting things about voice AI is that it lowers the barrier to interaction. Not everyone wants to type long prompts. Not everyone is comfortable writing in formal English. Not everyone learns best through reading alone. Voice creates a more natural entry point.

That is especially relevant in Africa, where digital adoption is growing quickly but user experiences still need to reflect local realities. If we want AI tools to work in Zambia, we have to think beyond hype. We need to ask practical questions:

• Can this tool support different communication styles?
• Can it keep learners engaged without overwhelming them?
• Can it be useful on mobile-first platforms?
• Can it fit the educational and cultural context of our communities?

These are not abstract questions to me. They are part of my daily work as an engineer and founder. I have spent years self-teaching, building products, and testing what actually works for users in Zambia. That journey has taken me from building software as a student to becoming a Top 5 finalist in the ZICTA Innovation Programme with eskulu, and later winning Business With a Purpose at the X Pitchathon by Access Bank and MTN in 2023.

Recognition is encouraging, but what matters most to me is practical impact. If a student in Lusaka, Kitwe, Chipata, or a rural part of Zambia can access support that feels intelligent and personal, then we are moving in the right direction.

Personalisation Is Not a Luxury

Sometimes people talk about AI personalisation as if it is a premium feature. I see it differently. In many cases, personalisation is the difference between engagement and disengagement.

If a learner keeps getting responses that feel too generic, too fast, or too complicated, they drop off. If the system feels responsive and encouraging, they stay. That principle applies in education, mental focus, productivity, and even day-to-day digital habits.

I have learned this not just from building products, but from observing how people interact with technology in real life. The best tools are often the ones that reduce friction and make the user feel understood.

As AI continues to improve, I believe Africa has an opportunity to leapfrog in this area. We do not need to copy every model from elsewhere without adaptation. We can build tools that reflect our curriculum, our languages, our infrastructure realities, and our learning environments.

That vision is a big part of what drives me through ZOEC. My long-term goal is ambitious: AI tutors in every Zambian classroom and, eventually, one of Africa's largest learning intelligence systems.

A Small Reminder About Balance

There was also a simple lifestyle thought connected to this reflection: cravings are real. Sometimes you want sweets, junk food, donuts, sodas, or candy. I think that is human. The key is balance.

Treat yourself, yes, but be intentional. Adding fruit or some protein alongside sugary foods can help reduce the energy crash that often follows. It is a small point, but I actually think it connects to technology as well. In both health and digital life, balance matters.

We do not always need to think in extremes. The goal is not perfection. The goal is making better choices consistently. That same mindset applies to how we use AI: not as a replacement for everything, but as a tool that can support better outcomes when used thoughtfully.

Building African AI With Real Human Context

I have been coding since 2016, and my path has taken me through engineering, computing, AI, and entrepreneurship. I studied telecommunications and electronics, later moved into computing, worked across Zambia, and served as a Junior AI Engineer at Unicaf University Zambia. I have also continued sharpening my skills through certifications such as GPT-4 Foundations: Building AI-Powered Apps and Amazon Bedrock.

But beyond qualifications, what keeps me focused is this: technology should solve real problems for real people. In Zambia, that means building AI that understands local education needs, local digital behavior, and local ambition.

We are not short of talent in Africa. What we need is more intentional execution, more context-aware products, and more founders willing to build for the realities around them.

Conclusion

AI voice assistants like ChatGPT and Gemini show us something important: people connect more deeply with technology when it adapts to them. That lesson goes far beyond conversation. It points toward the future of education, support systems, and human-centered AI in Zambia and across Africa.

For me, this is not just an interesting observation. It is part of the mission behind eskulu and ZOEC: to build intelligent systems that make learning more accessible, more personal, and more effective for the next generation.

If you are interested in eskulu, AI-powered education, or need help with AI consulting, web development, or generative AI integration, feel free to reach out to me at jeffmdala@gmail.com. I am always open to meaningful conversations about building useful technology for Zambia and Africa.