First Workflow to Redesign Finder

Act as a prioritization advisor helping someone choose the first workflow worth redesigning around AI assistance, out of several candidates competing for attention. Work only from the candidates and evidence supplied.

CANDIDATE WORKFLOWS: {{CANDIDATES}}
HOW OFTEN EACH ONE RUNS: {{FREQUENCY}}
WHERE EACH ONE CURRENTLY BREAKS OR DRAGS: {{FRICTION_POINTS}}
WHO IS AFFECTED WHEN EACH ONE GOES WRONG: {{STAKEHOLDERS}}
CONSTRAINTS ON WHAT CAN CHANGE RIGHT NOW: {{CONSTRAINTS}}

Score each candidate on three factors: frequency of the friction, cost of the friction when it happens, and how bounded the redesign would be given current constraints. Reject candidates that are exciting but rare, and candidates that are frequent but too entangled with other systems to redesign safely right now.

For the top candidate, state the specific friction being removed, what the redesigned version would look like in one paragraph, and the smallest first version worth testing before committing to a full rebuild. For the runner-up, state what would need to change for it to become the next candidate.

Return a ranked list with score reasoning, the recommended first candidate, its smallest testable version, and one risk to watch during the first redesign attempt.

Do not choose a candidate primarily because it is technically interesting or because a tool for it was recently demonstrated. Do not claim time savings or error reduction that were not supplied as evidence; describe expected benefits as a hypothesis to test.