Historical note: this chapter uses a verified Gemini archive record from July 4, 2025 plus preserved project-chat anchors from the same day. Credentials, server details and sensitive implementation information are omitted.
A large project, one basic question
By July the trading bot was no longer a tiny experiment. There were files, trading logic, execution attempts and an increasingly ambitious plan. Yet one preserved question captures the human reality better than any architecture diagram: how do I enter the virtual environment?
A Python virtual environment, usually called a venv, is simply a separate box for one project's Python packages. It lets the bot use the versions it needs without disturbing other software on the machine.
This was not an expert pretending to build from scratch. Patryk was genuinely learning while building, often one command at a time, with AI acting as teacher, coder and debugger.
Then the code broke
The same day's preserved project history shows bot_loop.py failing because of an indentation error. In Python, indentation is part of the language itself. A few spaces in the wrong place can stop the whole program before the trading logic even gets a chance to run.
The workflow was practical rather than elegant. Patryk repeatedly asked for complete corrected files instead of isolated fragments, then copied the replacement into the project and tried again. The AI could propose the fix, but Patryk was still the hands at the machine.
Running did not mean trading
After the correction, the bot finally started. Then came another small lesson: it reported that it was outside its trading hours.
That message was not a victory and it was not a failure. It showed that getting software to launch is only one layer. A trading system can be technically alive while doing absolutely nothing in the market.
The gap was the story
There is a temptation to rewrite beginnings once a project becomes complicated. July 4 is useful because it prevents that. The project could already contain ambitious automation while its builder still needed help with basic Python operations.
That gap between the size of the dream and the level of knowledge was not a contradiction. It was the working method. Patryk asked, copied, tested, brought the error back to the AI and tried again. The Swarm was not yet a polished research mind. It was still largely a teacher and code partner learning how to work with him.
Three days later the question would change. The bot could run. But was running actually producing anything useful?
AUR is a research project. Historical posts are not investment advice and do not promise future returns.