An Interview with Deborah Adetunji, Developer of the TUMAINI Prototype
At the heart of the TUMAINI chatbot lies a simple yet powerful goal, to support young people with disabilities through accessible technology tailored to their needs. Designed as part of our AI4SDG initiative, TUMAINI offers users a safe and supportive space to access career guidance, mental health resources, and tools for navigating the digital world.
The chatbot was piloted with young Persons with Disabilities in mind. It responds to queries related to mental well-being, digital skills, and employability, providing easy-to-understand, localized support. TUMAINI reflects a thoughtful approach to tech design, one that centers inclusion and lived experience from the start.
In this conversation, we speak to Deborah Adetunji, the developer who designed the TUMAINI prototype, about her experience building the tool, how she approached designing for accessibility, and why cultural relevance and memory in chatbot systems matter.
Q: What unique challenges or opportunities did you encounter when designing TUMAINI specifically for young people with disabilities in Nairobi’s informal settlements? How did these factors shape your development process?
Deborah: Most of the chatbots I’ve built before focused on basic functions like booking or answering FAQs, but TUMAINI was different. Here, the challenge was deeper, creating something that truly serves and empowers young people with disabilities, many of whom face layered barriers. It meant thinking beyond just technology and understanding their environment, the language they use, and the support they need. When Christopher Harrison (Next Step Foundation’s Executive Chairman) approached me with this idea, I felt it was more than just a project. It aligned with my personal goal of using tech not just for money, but for real growth and meaningful change. These realities shaped everything, from how the chatbot speaks to users, to what resources it offers, to making sure it’s accessible and easy to navigate.
Q: Accessibility is often treated as an add-on. How did you integrate accessibility and inclusivity from the very start of TUMAINI’s design, and why do you think this approach is crucial?
Deborah: Inclusion starts at the design table, not as an afterthought. I made sure accessibility was baked into every step, whether that meant choosing simple, clear language or designing an interface that’s easy to use even for people with different types of disabilities. You can’t just add accessibility later and expect it to work well. This project was a reminder to me why I code, to solve real problems and ensure no one is left behind in the future we’re building. That mindset shaped the entire development process.
Q: Cultural relevance can make or break the success of a digital tool. How did you ensure TUMAINI’s language, tone, and interface resonated with its users on a local level? What role did community insights play in this?
Deborah: I started with research and listening deeply. I asked questions about the audience, explored visual language, tone, and local context. It wasn’t just about translating content but making sure the chatbot felt familiar and trustworthy. What phrases make sense in everyday conversation? What color palettes feel welcoming? What interface layouts are easiest to navigate? Community insights were key, understanding what works and what doesn’t helped me create a tool people could relate to and rely on.
Q: TUMAINI’s ability to “remember” previous conversations is a powerful feature. What were some of the technical challenges in building this memory function, and how did you overcome them?
Deborah: Memory was a big technical focus. It was important that if a user returns after some time, they can pick up where they left off, with past conversations still accessible. This is something many large systems struggle with. Based on early feedback, I improved the memory function to handle that. TUMAINI is built on Bubble with API connectors, so it’s scalable and can integrate with other platforms, including AI tools like ChatGPT or Jacaranda Health’s UlizaLlama model. Balancing performance with scalability was challenging but critical to the user experience.
Q: Looking ahead, you mentioned voice interaction as a next step. What impact do you foresee voice features having on the chatbot’s accessibility and user experience?
Deborah: Voice chat is a natural next step for accessibility. Not everyone using TUMAINI can easily type, so enabling users to speak and receive audio responses would open the tool to a wider audience. It could make the experience more inclusive, especially for those with visual impairments or limited literacy. Voice also brings a more human feel to digital interaction, which fits Tumaini’s goal of being a supportive companion.
Q: Reflecting on your journey with TUMAINI, what lessons about inclusive tech design and the role of African developers in this space stand out most to you?
Deborah: Building TUMAINI reinforced that tech design must start with people, especially those often overlooked. Inclusion requires commitment and deep empathy. African developers bring essential perspectives because we live these realities and understand the nuances. I believe we’re uniquely positioned to lead innovation that truly meets local needs. This project reminded me why I code: to solve real problems and make sure no one is left out of the future we’re building.
What’s next for TUMAINI
We’re working on expanding its knowledge base and training it with even more inclusive content. There are plans to roll it out more widely and integrate it with other support systems, like mentorship programs or helplines.
Ultimately, TUMAINI is a proof of concept that says: Inclusive, African-led AI is possible. And it starts with listening.
