AI co-pilot
Seeing what the AI actually did
Every AI action surfaces the tool calls it made — auditable.
After auto-fill, MISSED resolve, conflict resolve, or quick-add, the result toast has a "Show reasoning (N steps)" disclosure. Expand it to see the exact tool calls (search_guards → update_shift → ...).
This makes the AI's decisions auditable retroactively. If a dispatcher asks "why did it pick Mike?", the trace shows whether the AI searched guards, checked availability, ran compliance — or just picked the first match.