Cheyenne Algorithms LLC
Engineering perspective. AI as a tool. Software built to last.
A retired professional software engineer walks into the App Store. No team. No game engine. No 3D asset files. Just Swift, Metal, and an AI neural network trained to write code. Here is what actually happened — and what it means for the future of software engineering.
Read post →I spent decades as a professional software engineer. I have written code in environments where correctness was not optional — where a wrong answer had consequences. I understand what it means to build something that works exactly as specified.
When I retired, I decided to build an iOS app. Not because I needed to prove something. Because I had an idea, the skills to execute it, and a new kind of tool I wanted to understand: an AI neural network trained to write code.
The result is Fascinator: Fidget Lab — a physics-based 3D fidget toy for iPhone and iPad, built entirely in Swift and Apple's Metal GPU framework. No traditional 3D asset files. No game engine. Every object — fifteen of them — is procedurally generated from mathematical first principles: geometry, surface normals, lighting, and physics simulation all computed in code. The app is available on the App Store.
I built it alone.
Here is what I need to be precise about, because the public conversation around AI and software engineering gets this wrong constantly: the AI did not write the app. I did.
An AI neural network, used as a coding partner, is exactly that — a partner. It can produce code rapidly. It can suggest implementations, catch syntax errors, and work through boilerplate. What it cannot do is know what to build, why to build it that way, or whether the output is correct for the problem at hand.
Every architectural decision in Fascinator was made by a human who had the experience to know what to ask — and the judgment to know when the answer was wrong.
The physics model required understanding quaternion rotation, angular momentum, and damping coefficients. The rendering pipeline required understanding how Metal shaders execute on GPU hardware. The AI could help implement each of these once I knew precisely what was needed. It could not determine what was needed.
This is the distinction that matters: AI amplifies experienced engineers. It does not replace them. An experienced engineer with AI tools can deliver what previously required a team. A novice with AI tools produces something that looks like software until it doesn't work.
I founded Cheyenne Algorithms LLC on this thesis. The economics of software development are changing — not because AI writes code, but because experienced engineers who know how to direct AI can now operate at a scale that was previously impossible for individuals.
I am one engineer. I shipped a complete commercial iOS application — real-time 3D physics, GPU rendering, App Store distribution — on a timeline and budget that would have been unthinkable without AI assistance. The quality and correctness of the result reflect decades of engineering judgment, not the AI's capability.
For defense, aerospace, and business software clients where precision and accountability matter: this is what AI-assisted engineering actually looks like when done by someone who knows what they are doing.
The AI is a tool. I am the engineer.
— Amanda Giarla · Founder, Cheyenne Algorithms LLC · Colorado Springs, CO