Why AI pilots die in the second quarter

The pattern

First quarter: enthusiasm, demo, applause. Second quarter: nobody knows who owns it, the input data changed, and the pilot quietly shuts down. I’ve seen it in large corporations and I see it in SMEs. It isn’t a technology problem.

Three decisions made upfront

1. A measurable business outcome. At Caja Piura the goal wasn’t “use AI” — it was reducing the time to disburse a loan. With that north star, an alternative AI-based scoring model made sense and could be measured: −90% in lead time.
2. A standardized process. At Real Plaza, before automating anything, we standardized 80% of back-office processes onto a unified platform. Automating a different process per department is automating chaos.
3. An owner with a budget. If the pilot depends on one enthusiast’s spare time, it dies the moment that person moves to another project.

Lean before AI

My training as a Lean Six Sigma Black Belt left me with a reflex: fix the flow first, then add the tool. AI amplifies what’s already there. If the process is clear, it speeds it up; if it’s confusing, it makes the confusion faster.