The field guide behind the method
The Decision Instrument
From Rough Intent to Finished Work
A 213-page operating guide for people who want to use AI with structure, evidence, judgment, and control—not just ask a blank box for a polished answer.
The Decision
Instrument
A Field Guide to AI Judgment, Workflows, and Trust
Gabriel Lacroix
Not a prompt formula
A way of moving through work with AI.
The book begins where the useful work actually begins: messy reality. It shows how to build a fuller field, find the decision inside it, make a concrete object, test that object, and preserve enough state to continue later.
The machine can move quickly once the field is built. The human work is deciding what belongs in that field—and what must remain under human judgment.
The sequence is the method
Five movements from rough intent to finished work.
- 01
Load the messy reality
Bring the records, events, constraints, unknowns, and consequences into the field before asking for an answer.
- 02
Turn reality into working context
Sort the material, preserve provenance, distinguish state from story, and keep rejected directions visible.
- 03
Find the decision
Name the operands, test competing frames, grade the evidence, and identify the next move without forcing certainty.
- 04
Make the work real
Turn the decision into a message, brief, file, workflow, prototype, or bounded automation that can survive outside the chat.
- 05
Verify and carry the state forward
Attack the object before reality does, preserve the source of truth, and return without starting the project over.
Inside the instrument
Fifty chapters. Five parts. One operating practice.
The guide moves through business analysis, research, writing, files, automation, project state, credibility, review, and recovery without separating AI from the real work it has to survive.
- Part 1Entering the work
- Part 2Loading reality
- Part 3Finding the move
- Part 4Making the work real
- Part 5Keeping the work alive
Use the method on real work