The 3-Check
Answers: How do I know this document is actually the thing I should be relying on right now?
An AI-sourced or AI-summarized document is about to inform a decision, a filing, or an answer to someone else.
Specimen 07Rev A · two rooms
The catalog · Ten instruments, one shape10 instruments · more as rooms produce them
Plate III · fig. 4
Identification key
Every entry, the same ten fields
The instruments vary; the shape does not. The mechanic names what structurally produces the insight — if you cannot state it in one line, it is not an instrument yet, it is a discussion topic. Came from carries the room it was born in, which is the part a competitor cannot fake.
Reading an entry
- Time
- A real number, not a range that means "we don't know".
- Group
- Solo · pairs · table · whole room.
- Rev
- How many different rooms it has survived. Three is the bar.
Rev
- B
- survived three different rooms — 4 in the field
- A
- survived two — 6 in the notebook
- Neither
- nothing enters the catalog on one success
Commonly mistaken for
- A maturity model
- The Scorecard
- A prompt engineering exercise
- What Goes In
- An efficiency metric
- The Outcome Test
- A status update
- The Pace Script
- Fact-checking
- The Three Verification Questions
- A process map
- Ten Steps to Three
- A citation check
- The 3-Check
- A skills assessment
- Find Your Rung
- A KPI review
- The Counterfeit Test
- A standard five-whys
- Name Which One It Was
Solo-first · Six you can run at your desk6 instruments · nothing needed but you
Find Your Rung
Answers: Where am I actually on the path from using AI to having a system, and what's the one next thing to do about it?
Someone wants to get better at AI with no sense of what that would concretely mean for them this quarter.
Specimen 08Rev A · two rooms
The Outcome Test
Answers: Do we actually have an outcome here, or just something that looks like one?
A team presents a win and nobody in the room can quite say who is better off because of it.
Specimen 03Rev B · three rooms
The Three Verification Questions
Answers: Can I actually trust this AI-produced assessment, or does it just sound authoritative?
A tool hands back a summary, a finding, or a risk assessment, and the instinct is to accept it because it reads clean.
Specimen 05Rev A · two rooms
The Counterfeit Test
Answers: Could this definition of success be satisfied by a team that did nothing of value?
A metric or a definition of done keeps getting hit and the room still isn't sure anything real happened.
Specimen 09Rev A · two rooms
The Pace Script
Answers: How do I defend our AI pace to the person who thinks we're too slow, and to the person who thinks we shipped too fast, without changing my story?
You're about to walk into a room that will push from one of those two directions and you don't have your three sentences ready.
Specimen 04Rev B · three rooms
With others · Four that need other people in the room4 instruments · a table, not a desk
The Scorecard
Answers: Which of our nine AI habits is actually working, and which one is quietly setting our pace?
Leadership wants an AI maturity number and nobody in the room trusts the one a vendor would hand them.
Specimen 01Rev B · three rooms
What Goes In
Answers: What actually belongs in the context we hand the model, and whose call is that?
A team keeps adding to a prompt or a project's context and the outputs are getting worse, not better, and nobody can explain why.
Specimen 02Rev B · three rooms
Ten Steps to Three
Answers: Why does our workflow have this many steps, and which ones actually produce the outcome we want?
A process has accumulated steps over time and nobody can explain what several of them are still for.
Specimen 06Rev A · two rooms
Name Which One It Was
Answers: When something went wrong with AI-assisted work, was it an augmentation failure, an acceleration failure, or an interpretation failure, and are we fixing the right one?
A post-mortem is underway and the instinct is to blame the AI as a single undifferentiated cause.
Specimen 10Rev A · two rooms