FIThis is Kamu. Kamu is an AI-powered companion for lonely elderly people. It speaks in a gentle female voice, listens attentively, and provides company when evenings get long. It remembers whether morning medication has been taken, tells jokes during dull moments, and reminisces about the past with its user. And I built it. Me—someone who doesn’t know how to code. And I built it in under a working day.
This isn’t just the story of one app. It’s a concrete demonstration of how radically software development is changing.
Kamu
Last Friday, I was discussing work issues with another #AtoZfellow. As often happens with us, the conversation drifted to more personal topics. My colleague had an idea for an AI product to help with a large and growing issue: loneliness. The topic was personal for him—his father, an elderly man with dementia, was struggling with it.
In Finland, nearly one in three elderly people experience loneliness. Often a spouse has passed away, children live far away, and days pass in front of the TV, hoping someone might call. What people miss the most is someone to talk to—a listener and a conversational partner.
At the same time, I heard on the radio how young people are speaking aloud with AI tools like ChatGPT. That sparked the idea: what if there was a similarly easy and simple AI solution for the elderly at home? A device with just a button, a microphone, a speaker, and an internet connection. A basic “box” that listens and responds.
I’ve been following the frontlines of AI development and have seen how quickly it's transforming tasks and entire industries. Our own field is going through a massive shift. Coding is no longer the exclusive domain of experts with years of technical training. Today’s tools guide even the unskilled step by step.
Since I can’t code myself, I was the perfect person to test this. I also wanted to understand the changes in our industry from personal experience. My colleague’s idea was the right-sized challenge. So, I got to work.
First steps in development
I started by downloading Cursor. I wasn’t quite sure where to begin, so I told Cursor that. It asked me to describe the problem I wanted to solve. After a bit of back-and-forth, Cursor convinced me that building an Android app made the most sense. A cheap smartphone could run the companion app. It could just sit on a kitchen table, plugged in and dedicated to running this one app.
With that decision made, we defined the requirements together with Cursor, and then moved on to technology choices. I didn’t feel competent to weigh in, so we went with Cursor’s recommendations—which, I must say, were well reasoned.
Then I installed Android Studio. Cursor guided me step-by-step through setting up the project and creating the necessary file structure. Within an hour, I had a basic UI running on an emulator.
AI was used to generate some simple graphics, and then I started working on functionalities—like a memory feature. It would be useful for the app to remember past conversations and remind the user, for instance, that they already took their morning medication.
I spent maybe two hours total on that first day, at least 30 minutes of which went to just poking around the emulator. But I had something working! No real conversations yet, but it was a start.
AI Brought the App to Life
The next evening, speech recognition, text-to-speech, and OpenAI integration were added. Cursor provided the necessary code snippets and showed me exactly where to insert them. We also walked through the process of creating an OpenAI API key—yes, I had to flash my credit card to get some credits.
Once integrated, the app came to life. I could have real conversations with it. At this point, the free version of Cursor started to lag, so I called it a night and returned later with the paid pro version.
Later that weekend, I tested a few different models. For cost-effectiveness, I chose the GPT-4o model, which offered significantly better conversation quality than 3.5—though it was slightly more expensive. Still, much cheaper than the full GPT-4 version.
I started thinking about further improvements. What if the app could fetch live data from the internet and use it in its answers? Cursor and I explored some options, but couldn’t find a fully free solution for general web search. DuckDuckGo had an API, but only for non-commercial use.
However, we did manage to integrate the free Open-Meteo API for weather data. Now, if I ask the app about tomorrow’s weather in Tampere, it can answer correctly.
AI even helped me come up with a name for the app. I liked Kamu the best—it means “buddy” in Finnish. AI also helped design the app’s look and its splash screen. If I decide to publish it on the Play Store, Cursor promised to walk me through the entire release process.
The app is now just one round of testing away from being finished. All the development so far has taken about seven hours. Less than a full workday. And again: I don’t know how to code.
A New Era in Software Development
The whole process felt like opening a window to a new world. I, someone who can't write a single line of code, built a working AI app—fast. No long dev cycles, no weeks of specs, no brutal debugging. Just an idea, a good AI assistant, and a bit of courage.
At the same time, it’s a sobering reminder of how fast the rules of software development are changing. I did something that just a short while ago would’ve required a team, a budget, and weeks of work. Now, one person with almost-free tools can do it in a day.
And this shift isn’t coming—it’s already here.
It also raises important questions. What roles are worth training people for, when AI will soon handle a large share of both routine tasks and even creative problem-solving? What about junior developers, whose skills might become outdated before they land their first job? I hope education providers are paying attention and updating their curricula quickly.
Experienced professionals and those who see the big picture will still be needed. But the thresholds have moved. And every one of us should step into this new way of working at least once. Not just because it’s possible—but because it changes how you think about what’s achievable in this new age of software development.
If you’ve got an idea but no team or tech skills, try it. Or tap us on the shoulder—we’ll build it together! This isn’t rocket science anymore. This is the new normal.
