Scoping a pilot that survives review
Automation pilots rarely die of bad technology. They die because the exception path was never agreed, and month three is far too late to negotiate it.
Yusuf Kaya
Pilots rarely fail on capability. They fail because nobody agreed in advance what the agent was permitted to get wrong, so the first visible mistake becomes a referendum on the whole programme.
Three things to settle before you commit
First, the confidence floor: the score below which the agent does not act. Second, the escalation route: which human queue sub-floor work lands in, and who watches it. Third, the review cadence: when the exception log is read together, and whether the floor moves.
None of it is technical. All of it is the difference between a pilot that becomes infrastructure and one that quietly stops being used while still costing money every month.