The Pilot: Twelve Weeks to Prove the Assistant Works

Twelve weeks to prove an AI assistant works for a bank's fees and contracts desk, and a dashboard that reports the savings before the mistakes. The participant leads the pilot at Banco Serrella, a regional bank in Valencia: eight specialists, six task types that come in look-alike pairs, and a COO who will ask two questions on 2 April 2027 - should we roll it out, and what is it worth?

Over three four-week sprints the participant sets, task by task, whether the assistant works autonomously, under spot-check, under full review or not at all; orders audits, 50-item samples whose results arrive a sprint later; and makes six hands-on checks of real desk work, each with a stated confidence. A model update in week 9 moves two tasks across the frontier in opposite directions.

Errors that reach clients land 5 to 10 weeks late, so the dashboard rewards the riskiest policy for two months. At go / no-go the participant chooses a rollout and reports the pilot's ROI; the week-16 reveal shows the true value, the errors still on their way, the frontier map and a profile.

The score weighs pilot value, rollout value, calibration of the six calls, exploration before scaling, adaptation after the update and an honest ROI report. The Express front door of the AI-adoption flagship line: individual play, about 25 minutes, English and Spanish.

Una simulación avanzada para participantes acostumbrados a trabajar con los principales marcos y dilemas de la materia, pensada para formación de directivos y profesionales con experiencia.