When the App Said 'Wait' — And How I Found the Right Question for Local Dialect
Building Local Dialect, a language learning app for Kenyan and African languages, taught me the value of patience and the power of asking the right questions — especially when the AI didn’t quite know the answer.
"When the App Said 'Wait' — And How I Found the Right Question for Local Dialect Building Local Dialect, a language learning app for Kenyan and African languages, taught me the value of patience and the power of asking the right questions — especially when the AI didn’t quite know the answer. There’s a strange kind of quiet that comes with building something that matters — something that has never been built before. For Local Dialect, that quiet was louder than any code I’d ever written. It wasn’t just about writing an app; it was about creating a space where people could reconnect with their heritage languages, the ones that live in the rhythm of market chatter and the lilt of a grandmother’s voice. And that, I quickly learned, was not a task AI could solve alone.
At first, I thought the AI would help me structure the lessons. I asked it to design a curriculum for Swahili and other Kenyan dialects, expecting a neat, step-by-step plan. But the AI kept coming back with something unexpected: a question. ‘What is the goal of the learner?’ it asked. Then, ‘What do you want them to feel when they finish?’ I had no answers. I had only a vision — and it wasn’t enough.
That was the first hard moment. I had to go back to the people I was building this for. I asked teachers, parents, and learners. I listened to their stories and the way they spoke about language. What I found was not a rigid curriculum, but a living, breathing connection to identity. And that connection was something no AI could generate — it had to be discovered.
The second hard moment came when I tried to build the app itself. The tools I used were not made for low-resource languages. There was no training data, no standard fonts, no voice recognition models that could understand the regional dialects. I had to find workarounds. I built custom voice recognition models using open-source tools. I designed the UI with accessibility in mind, making sure the app could be used even on older phones. And every time I hit a wall, I turned back to the AI — not to ask for a solution, but to ask the right question.
Then came the surprise. The AI, which had been a tool for generating ideas and code, became something else. It started to understand the cultural nuances I was trying to build into the app. It suggested ways to phrase lessons that felt authentic, not just educational. It even helped me find resources I hadn’t known existed — audio recordings of elders speaking in dialects I’d never heard before. And that, I realized, was the most surprising part: the AI wasn’t just a tool. It was a collaborator, helping me find the right path when I didn’t know where to go.
What I’m proud of is that Local Dialect is not just a product. It’s a conversation — between me, the AI, and the people who speak these languages. It’s a reminder that building something real, something meaningful, often requires more than just code. It requires listening, asking the right questions, and being willing to wait for the answer.
I’m still learning, still building, still asking questions. But now, I know that the hardest parts of the journey are not
