Founder, One Prompt Engineering
One prompt.
Infinite possibilities.
21, based in Georgia. I build complex things out of carefully engineered prompts — instructions with enough structure, context and constraint that the result stops being a guess and starts being a specification.
Build a complete solution for this idea.
One prompt, one complete result.
A demonstration of the method, not a live model.
/about
I'm Sebastian — most people call me seba. I'm 21, based in Georgia, and I spend most of my time on one question: how far does a model actually go when the instruction is built properly?
Not tricks or magic words. Structure. An objective, the context a model would otherwise guess at, the constraints it must not cross, the logic it should follow, and a clear definition of what finished looks like — assembled before a single token is generated.
I'm a student. I build software. And I treat prompting as an engineering discipline rather than a conversation, because the distance between a mediocre answer and a finished system is almost never the model.
/philosophy
A chatbot answers questions. An engineered prompt defines a process — it carries the objective, the context, the constraints, the logic and the shape of the expected result, so a system has everything it needs to arrive somewhere on purpose.
Written that way, an instruction stops being a question and becomes executable.
A rough intention. Valuable, ambitious, and completely unactionable to a machine in the state it usually arrives in.
The idea rewritten as a specification: what is being built, who it is for, and what counts as finished.
Everything the model would otherwise invent: source material, rules, order of operations, and the limits it is not allowed to cross.
Execution. The model is not being asked to be creative about the objective — only about the path to it.
One pass, one deliverable, measured against the criteria that were written down before it started.
/demonstration
Pick a brief, run it, and watch a single instruction resolve into a finished result.
Build a complete solution for this idea.
One prompt, one complete result.
A front-end demonstration. No model is called.
/work
Open an entry to see how the brief was specified. Entries marked example are placeholders, kept until the real build ships.
The method itself — a repeatable structure for turning a one-line brief into a fully specified build.
Objective, context, constraints, logic and acceptance criteria, written in a fixed order so the same brief produces the same class of result every time.
A versioned collection of prompt specifications, each stored with the constraints and output contract it was written against.
Every entry keeps the brief, the finished instruction and the result it was accepted on, so a working prompt gets reused instead of rewritten.
Repetitive work described once and executed on a schedule — where the prompt is the program.
A brief is written, tested against edge cases, then handed to a runner that executes it unsupervised and reports what it changed.
Designed, written, built and deployed from a single engineered prompt.
Static front end, no framework, no runtime dependencies, deployed on Cloudflare Pages. The build is the argument the site is making.
/founder
One Prompt Engineering is the name for how I work: complex problems get simpler the moment the right context, structure and instructions are in place. Most of the effort happens before anything is generated.
It isn't a claim that one sentence can solve anything. It's a claim about where the leverage sits — and the leverage is almost always in the instruction, not the model.
/skills
Areas of practice and active interest — the tools the method gets applied through.
/journey
Started exploring technology and AI, mostly by taking things apart to see what they would do.
Began pushing AI past ordinary chatbot use — longer chains, harder tasks, stranger requests.
Developed increasingly structured techniques, and noticed results tracked the structure, not the wording.
Turned the method into a philosophy, and the philosophy into One Prompt Engineering.
Continuing to experiment, build, and find the edge of what a single well-built instruction can accomplish.
/contact
Give me the problem. Let's see what one prompt can do.
Click to copy the handle. Send the problem, and let's see what one prompt does with it.