Pablo Hernandez Leal

CTO at MultiplAI

I build AI systems for regulated financial workflows and write about applied AI, decision-making, experimentation, and the problems that appear when technology meets the real world.

What I work on

AI systems in regulated environments

Building for accuracy, evidence, traceability, and human review.

From research to production

Turning experimental findings into engineering decisions under operational constraints.

Technical decisions and teams

Working across engineering, product, and domain expertise when requirements or evidence are incomplete.

Selected writing

Sometimes You Need Kirk. Sometimes You Need Spock.Sep 2026When should a decision preserve options, and when is it time to commit? The Executive Summary Is Not a SummaryOct 2026An abstract says what the work found. An executive summary explains what it means for the decision. The Metric Can Be Right and the Decision Can Still Be WrongSep 2026Why a correct metric can still support the wrong decision when it measures only part of the system. View all writing →

Interactive experiments

View all experiments →

Small models you can change to see how a decision unfolds.

Build · test · stop

Two Clocks

See how learning can improve a decision while the opportunity to act narrows.

Model · errors · review

The Metric Can Be Right

Choose a fraud-detection model, then reveal what its accuracy score leaves out.

Decision tools

A toolbox for unclear decisions

Eighteen questions drawn from research, startups, engineering, games, and experience. They are not rules—just different lenses worth trying when the answer is unclear.

Explore the toolbox →