How I build
I built dashdahlberg.com using AI development systems that pair large language models with agentic harnesses for context management, tool use, code execution, and iterative feedback. I use these systems where they are allowed because they help me move from an idea to something I can test, use, and improve.
The short version
My academic work is still evaluated in settings where AI is not part of the work: SAT and AP exams, in-class tests, and assignments where my school does not allow AI. Those results are mine.
Outside the classroom, I use AI as a building tool. It helped me create this website, an orbital modeling project for exploring astrophysics, a home run simulator that connects physics and baseball, and customer-support tools for Edge of the Wedge. Those results are also mine, but I have used an expanded set of tools to produce them.
The important part for me is not that AI writes code. It is that AI lets me ask better questions, test assumptions, revise the product, and build things I would not have been able to build as quickly on my own.
What AI helped with
AI-assisted coding tools helped me create the structure, pages, components, styling, and iteration loop for this site. I can describe what I want a page to do, review a working version, notice what feels wrong, and keep refining it until the result matches the idea.
AI helped me build interactive tools so I could explore orbital mechanics and baseball flight models instead of only reading about them. The model gives me something concrete to question: I can change an input, compare the result with the physics, and find the assumptions that need more work.
AI helped me think through faster customer answers and workflow support for Edge of the Wedge. The goal is not to make a generic chatbot. It is to understand the real questions customers ask, give useful answers, and make the business easier to run.