An agent gets a business task and starts opening files. A controller points to a service, the service leads somewhere else, and understanding the operation becomes a hunt for clues. I recognised the investigation: I had been doing it myself for years.

This talk follows what happened when I started looking at architecture through the eyes of that reader. We examine how an operation is scattered across a codebase, what becomes easier when its story stays together, and which abstractions still earn their place. The approach grew from a colleague’s work on making software understandable to people joining the team. Coding agents gave us another reason to examine it.

Python examples and recorded demonstrations let us compare structures, distinguish repeated code from duplicated knowledge, and explore the trade-offs. You leave with a way to investigate your own codebase and decide where a clearer business story would help the next reader, human or agent.

For developers and architects interested in how code structure affects understanding. This session expands the investigation behind “Your codebase is the actual prompt” through examples and architectural trade-offs. Each session stands alone; organisers can choose the broader discovery or this deeper exploration. Python makes the choices visible, while the discussion applies across stacks.
Develhope recording in Italian: https://youtu.be/VXqBT6jueKA
Example code: https://github.com/lazyoft/building-with-raymond