Building Centrigon — One Piece at a Time
Shipping the first small piece of a much bigger system
Every time an African developer talks about building a global tech product, the same criticism eventually appears.
“Why not focus on African problems instead?”
“AI startups from Africa are unrealistic.”
“You’re competing with companies that have billions of dollars.”
To be fair, it’s not a completely unreasonable reaction. The companies pushing the boundaries of modern AI today—like OpenAI—operate with enormous research teams, infrastructure, and funding that runs into the billions.
So when someone building from Africa says they’re working on something ambitious, the default assumption is that they’re either naïve or wildly optimistic.
For what it’s worth, I’ve heard those criticisms more than once.
But that’s partly how Centrigon started.
The Moonshot
Centrigon isn’t meant to be just another app.
The long-term vision is a system that is:
- multimodal
- persistent
- autonomous
- decentralized
In other words, an intelligence layer that doesn’t simply respond to prompts, but can operate continuously, interact across different forms of data, and evolve over time.
That’s a big idea. Possibly an unreasonable one.
The only way I know how to approach a project that large is to start with something small and keep going.
The Reality of Building Big Systems
Ambitious projects have an inconvenient requirement: time.
And if you want to stay in the game long enough to build something ambitious, you need a few things along the way:
- infrastructure
- users
- distribution
- and ideally some form of sustainability
You can’t disappear for five years building a moonshot and hope the world is still waiting when you return.
So instead of trying to build the entire system in isolation, I decided to approach it differently.
Build it piece by piece.
Building It Piece by Piece
Centrigon is a large system.
The challenge with building something like that is that most of its capabilities only become visible once everything exists together. If you try to build the whole thing at once, you can spend years working before anyone ever sees anything.
I wanted to take a different approach.
Instead of waiting for the full system to exist, I’m starting to release individual capabilities of Centrigon as standalone tools.
Each one is small. Each one does a single job. But each one also represents a real piece of the larger architecture.
Think of them as early fragments of the system — released one at a time.
The First Piece
The first piece I decided to release publicly is an image converter.
On the surface it’s simple: convert images between common formats like PNG, JPG, and WEBP.
But under the hood, it’s also a small step toward something larger.
Image processing is one of the building blocks of computer vision, which forms part of the broader capabilities Centrigon will eventually need.
So while this tool is small, it represents one of the components the larger system will rely on.
Another design choice I liked while building it: the conversion runs directly in the browser. That means images never leave the user’s device.
Simple, fast, and private.
If you’re curious, you can try it here:
https://centrigon.com/tools/image-converter
The Experiment
The bigger experiment behind this isn’t the converter itself.
It’s a question.
Can someone building from Africa ship useful pieces of technology that people around the world actually use?
If the answer is yes, then something interesting starts to happen.
Each small piece becomes a building block.
And over time, those pieces start forming something much larger.
The Long Road
Centrigon is still the destination.
But systems like that rarely appear fully formed. They grow gradually from smaller components that prove themselves in the real world.
So for now, the plan is simple:
- Ship something useful.
- Learn from how people use it.
- Then build the next piece.
For now, this converter is just one small piece of Centrigon — but every system has to start somewhere.
Comments