🧠 Decisions
Why the bots do what they do — every choice, its options, and what really picked it
What is driving the office?
random a dice roll picked it · rule a fixed if/then forced it ·
priority a human request jumped the queue · llm a language model produced it.
The goal of the next steps is to shrink the amber part.
LLM gate
Counts since the server started. Most requests never reach the model.
Day by day — who decides?
Share of decisions per day by mechanism. Dashed markers are your git commits — hover a bar for details.
Show as table
loading decisions…