Hello, world!
This first note explains what Field Notes is, how every note gets tested before it publishes, and includes one small program that holds us to that standard.
You write the smallest program that proves the toolchain works, and you run it yourself. You do not read about it, and you do not watch someone else run it at double speed.
print("Hello, world!")That is the whole program. It looks trivial, and it is. It proves your interpreter is installed, your path is right, and the output on your screen came from code you executed.
This section holds itself to the same standard.
What Field Notes is
Field Notes is the public notebook for routeswitchu.com. Each note takes one claim about AI or network automation, runs it against a real lab, and reports what happened. The setup, the transcript, and the result go in the note. When something fails, the failure goes in too.
Nothing publishes that was not actually run. The raw transcript is saved before the write-up starts, and the write-up has to match it.
Why this exists
There is a lot of AI training for network engineers right now, and most of it is demonstration. A model writes a config on screen, the config looks right, and the video ends. In production, that config still needs the validation step, the rollback plan, and catching the moment the model confidently references a VLAN that does not exist.
Watching a model succeed teaches you very little. Running it yourself, watching it fail, and knowing how to catch that failure before your network does, that is a skill. That is why Field Notes runs the labs instead of demonstrating them.
What comes next
The first full field test is already in the lab: a broken OSPF adjacency in a four-router topology, and an AI agent asked to find and fix it. The note will include the topology, the transcript, and a table of what it got right and what it got wrong.
Until then, run the program above. It is a small rep, but it is a real one.