Just Because We Can Doesn't Mean We Should
When I started building Tonely AI, the first question was how to create a real solution. The second question mattered more: how to do it without losing my soul.
When I started building Tonely AI, the first two questions that came to mind were: how can I create a solution for an actual problem — people communicating in harmful ways digitally — and how can I do it without losing my soul?
The second question was more important to me than anything.
As a Christian, I hold strong beliefs. There's a verse in the Bible that basically says, what's the point of gaining the whole world but losing your soul? I really wanted to create something that could make a difference, but I didn't want to lose myself in the process.
I wanted Tonely to be available to everyone, not be intrusive, and actually do what it's meant to do. I'm British with Nigerian descent, which is a very funny combination. As a Brit, there's always that underlying feeling of 'it could be better'. As a Nigerian, there's that underlying perfectionism — if it's not perfect, it shouldn't be shared. So as I battled those personal tendencies, I had to accept one thing: I was going to have to release something that isn't perfect.
At the time, Tonely was around 89% accurate. Today it's around 93%, with some categories reaching 99%.
What I Could Have Built
One thing I discovered while building Tonely is this: technically, if I had wanted to and if I was a psychotic, money-hungry person, I could have built something far more powerful with the data I had created.
- I could have made Tonely respond personally to you, giving it more autonomy to develop a personality around your communication.
- I could have made it write for you, in your own style, while sounding more neutral.
- I could have made it become smarter and smarter with every conversation you had.
But the more I researched AI, the more uncomfortable I became. Not because the technology isn't impressive. Because of the incentives.
The Problem With Incentives
Many AI systems become better because people unknowingly teach them. Every prompt, every correction, every conversation, every mistake — it all becomes another opportunity for the model to learn. People are unknowingly donating their thoughts, their writing styles, their humour, their emotions, their fears and sometimes their most vulnerable moments. In many cases, they have no idea just how valuable that information is.
They're effectively working for billion-dollar companies without ever receiving a payslip. They're becoming unpaid interns.
That didn't sit right with me. I wanted Tonely to become better because I made it better — not because my users unknowingly became part of the workforce. So I made the difficult decision to keep everything on-device.
That decision makes my life much harder. It means I don't have millions of conversations flowing back to me every day to improve the model. It means I have to keep creating data, testing, refining and retraining the system myself. But I'd rather work harder than ask my users to quietly work for me.
The Question We've Forgotten to Ask
Tonely can currently detect over 25 communication categories. Everything runs privately on your device, so your conversations never leave it. If I could put those guard rails in place, then the larger AI companies can too.
The AI industry spends a lot of time asking: 'Can we build this?' I think we've forgotten to ask another question: 'Should we?' They're two very different questions.
As engineers, we naturally love solving difficult problems. As founders, we naturally want growth. As businesses, we naturally want profit. None of those things are wrong. But if profit becomes the thing that drives every decision, we'll eventually build technology that serves companies better than it serves people.
If an AI company makes more money the longer you stay on its platform, the goal quietly shifts from serving you — to keeping you. That's a subtle difference. But I think it's a dangerous one.
What Drives My Decisions
One day I'll stand before God. He isn't going to ask me how many downloads Tonely had, how much money I made, or how many users I managed to collect. He'll ask what I did with what He entrusted to me. That changes how I build. It changes the decisions I make. It changes the features I refuse to build.
I care far more about my soul than I do about building the biggest AI company in the world. If I ever have to choose between the two — my soul wins. Every single time.