Most writing about automation is produced by the people who profit when there is more of it. The vendor wants the scope to grow. The platform wants the seat count to rise. The consultancy wants the programme to run for another year. None of them are positioned to tell a leader the one thing that often matters most, which is where automation should stop.
This journal starts from the opposite place. The work I do is counsel, not software. I have nothing to sell except the position a client can hold afterward, the document they can defend in a board meeting, and the name of the human who stays responsible when a decision is questioned.
So these notes are about the boundary. Where human judgment belongs. Where an agent can help while a person keeps the final call. Where quiet automation is the right answer, and where it quietly erodes the very thing a brand was trusted for. The interesting question is almost never the tool. It is the line.
What you will find here
Short pieces, in English, published on this site first. Some will be drawn from conversations I cannot name. Some will take apart a public failure that everyone already half understands. All of them will try to leave you with a sharper way to decide, not a louder opinion to repeat.
One standard runs through all of it
Where consequence is high, a named human stays responsible.
That is the whole practice in a sentence. If a process can put money, trust, or a person at risk, someone should be able to say who answered for it. Most operations cannot, yet. These notes are an argument for fixing that before a regulator, a board, or a client asks the question for you.
If a piece here resembles a situation you are sitting in, that is usually a good moment to begin a conversation.