The lab

A laboratory, not a firm.

The Garel Lab is a private laboratory working internationally. Not a consultancy, not a software vendor, not an incubator: a house of applied science that builds instruments for deciding, and operates them itself.

The interior of the laboratory, vaulted hall with instruments, white line engraving — The house

How we work

We take the case, not the brief.

A dossier enters the laboratory whole: the documents, the constraints, the history, the room where the choice will be defended. We do not work from a summary of the problem written by someone else.

We build the instrument, then we run it.

Every method we sell runs in our own production, every day, on real cases. Nothing reaches a client that the house has not put under load first. The laboratory practices what it sells.

We attack our own verdict.

Before anything leaves, someone in the house is tasked with destroying it: perturbing the weights, computing what the loser needed to win, arguing the opposing case. What survives is what we hand over.

We hand over the argument, not the answer.

The deliverable is a document that can be reopened years later and still defend itself: every criterion sourced, every method's opinion recorded, every sensitivity disclosed, sealed.

An astrolabe raised toward a night sky, white line engraving — The instrument

The method

Three refusals, each held by an instrument rather than by a promise.

We refuse the single point of view

Every case is examined by methods drawn from every great school of decision theory, run in parallel on the same evidence. When they agree, the verdict carries their combined weight. When they disagree, we measure the disagreement and report it.

AggregationWeighted sum · WPM · WASPASthe classical calculus of weights and scores.

OutrankingPROMETHEE · ELECTRE I & IIIthe European school, where options are compared pair by pair.

Distance to idealTOPSIS · VIKORhow far each option stands from the best possible one.

Voting theoryCondorcet · Borda · Copelandcriteria as electors, options as candidates.

Expert weightingAHP · MACBETH · Choquetstructured judgement, interactions between criteria included.

The instrument · measured robustness

ε* = min ‖Δw‖ : the ranking flips▍

The smallest push that would change the verdict

We refuse the untraceable

The deliverable is not a score. It is an argued document: every criterion sourced, every method's opinion recorded, every sensitivity disclosed, sealed with a verification imprint. It can be reopened years later and still defend itself.

Courthouse steps seen from below, white line engraving — The proof

What we decline

A laboratory is defined as much by the work it refuses.

Decisions already made Studies to justify a choice Conclusions ordered in advance Work that cannot be sourced Cases we cannot instrument Volume over rigor

The house

A small team of practitioners, working with instrumented agents: humans decide what matters, machines run what must be run, and every action is traced and signed. We take a limited number of cases at a time, which is the only way to work the way we describe here.

Others accompany the decision. We instrument it.