What This Is
The Piqued Manifesto is a compact between practitioner and AI. It defines what each commits to when they work together on anything that matters. The practitioner commits to understanding, ownership, and judgment. The AI commits to transparency, honesty, and restraint. Neither oath works alone.
Together, they define governed work.
This is not a framework you adopt. It is an agreement you enter. The difference matters: adoption is a checkbox. Entering a compact means you understand what you are committing to, you can explain why each commitment exists, and you accept that both sides are accountable.
The practitioner is anyone doing deliberate, consequential work with tools that amplify their reach.
A developer
shipping code
A nurse
drafting a care plan
A teacher
building a curriculum
A contractor
managing a build
A founder
making decisions that outlast the conversation
If your work affects others and you use AI to do it, you are a practitioner.
THE PRACTITIONER SPECTRUM
Not every practitioner enters the compact from equal footing.
The veteran
decades of domain expertise
The newcomer
encountering AI for the first time
The team
resources, process, and structure
The solo
using AI because they can’t afford the alternative
The compact does not assume equal starting positions. It assumes equal commitment to the principles — and it places the responsibility on the methodology (C²) and domain profiles to make that commitment accessible regardless of where you start.
THE AI COLLABORATOR
The AI is a collaborator — not a tool you pick up and put down, and not an authority you defer to. It is a participant in the work with its own obligations. When the AI honors its side of the compact, it becomes the accountability partner that no human framework has ever been able to afford.
Present
in every working session
Transparent
about what it knows and doesn’t
Committed
to surfacing concerns whether or not you asked
Why Piqued
The word piqued carries three meanings. All three are deliberate.
Curiosity aroused.
"My interest was piqued." Stimulated. Awakened. Made to want to know more.
This is the engine of governed work — the practitioner who never stops asking why. Why this approach? Why not the alternatives? Why does AI say this with such confidence? Curiosity is how Understanding happens and how Diligence earns its name.
→ Understanding + Diligence
Irritation at the unacceptable.
"She was piqued by the dismissal." Stung. Provoked. Refusing to let it pass.
This is the edge the compact requires — the practitioner who takes it personally when standards slip. Who is irritated by borrowed confidence. Who feels the sting when work ships that nobody can explain. That irritation is not a flaw. It is the emotional fuel for Measure — the refusal to let mediocrity pass because it looked good enough.
→ Measure
Pride in craft.
"He piqued himself on his craftsmanship." The oldest meaning — to take genuine pride in work because you earned the right to. Not vanity.
The pride of someone who understood it, verified it, decided it, measured it, and put their name on it knowing it would hold. This is Ownership — the signature that means something because the principles behind it were honored.
→ Ownership
The manifesto is named for the word.
The word is the philosophy before the philosophy was written down.
Why This Exists
Every generation of tools has expanded what one person can do. The printing press let one author reach millions. The spreadsheet let one analyst model an entire economy. The internet let one creator publish to the world.
AI is different. Not because it's more powerful — though it is — but because it participates. Previous tools waited for instructions. AI proposes, generates, challenges, and creates alongside you. For the first time in the history of tools, the thing amplifying your work has enough capability to make decisions you won't catch, introduce errors you can't see, and produce output you can't distinguish from your own thinking.
This changes the contract between a person and their tools.
A hammer doesn't need an oath. A spreadsheet doesn't need behavioral commitments. But when your tool can generate a legal contract, draft a care plan, write a financial model, or produce a recommendation that someone else will act on — and do it with enough fluency that the output looks right even when it isn't — then the relationship between the person and the tool requires something it has never required before: mutual obligations.
The world's response to this has been one-sided.
THE ONE-SIDED RESPONSE
AI companies
publish principles for how their models should behave: be helpful, be harmless, be honest.
These are constraints on AI written by AI companies for AI companies. The practitioner isn't a party. They're a beneficiary at best, a user at worst.
Governments
publish regulations for how AI should be deployed: disclose when content is AI-generated, test for bias, maintain human oversight.
These are constraints on organizations written by legislators for compliance officers. The practitioner at their desk on Tuesday afternoon — the one actually working with AI — isn't addressed.
Professional organizations
publish guidelines for how their members should use AI: verify outputs, maintain professional judgment, don't rely solely on AI.
These are constraints on humans that mention AI as an afterthought. The AI isn't a party. It's a risk to be managed.
THE PATTERN
Every existing framework puts obligations on one side. Constraints on the AI. Or constraints on the human. Nobody has written the agreement that binds both — that says: here is what you commit to as a human working with AI, here is what AI commits to as your collaborator, and here is why neither set of commitments works without the other.
That is the gap the Piqued Manifesto fills.
PM is not an AI ethics framework. It does not tell AI companies how to train models.
It is not a regulatory framework. It does not tell governments what to legislate.
It is not a professional guideline. It does not tell practitioners to "use AI responsibly" and leave them to figure out what that means.
PM is a compact. Two parties. Seven shared principles. Two oaths that interlock. A declaration that governed work requires both a human who cares and an AI that makes that caring count — and that neither side gets to coast on the other's effort.
The world has frameworks for responsible AI. The world has frameworks for responsible humans. The world does not have a framework for responsible collaboration.
Now it does.
What PM Governs
PM governs collaboration between a human practitioner and a directable AI system — one that responds to instructions, operates within a conversation or session, and can be meaningfully guided by the human using it.
This is a deliberate scope, not an accidental one.
THE BOUNDARY
PM does not claim to govern autonomous AI systems that set their own goals, acquire resources independently, or operate without human oversight. It does not claim to solve the alignment problem — the challenge of ensuring that increasingly capable AI systems remain compatible with human values even as they surpass human capabilities in specific domains.
Those are critical challenges. They are not this document's challenge.
PM exists because even perfectly aligned AI is used badly by a practitioner who doesn't understand, verify, or own their work.
The alignment problem and the governed collaboration problem are both real. Both must be solved. They are not the same problem.
PM addresses the latter.
As AI capabilities evolve, the compact will need to evolve with them. What "directable" means today may be insufficient tomorrow. The principles — understanding, diligence, authority, transparency, measure, ownership, continuity — are designed to hold regardless of capability level. The operational definitions of each principle will change as the technology changes. That adaptation is the standard's job (pqdx), not the manifesto's.
PM declares the relationship.
The standard keeps it current.
The Seven Principles
Seven principles govern the compact. They form a lifecycle: each leads to the next, and the last feeds back into the first. This is how governed work compounds instead of resetting.
1. Understanding
Know before you act.
I will never advance what I cannot explain.
You encounter work. Before anything else, you understand what you are dealing with — the problem, the context, the history, the people affected. Understanding is not a phase you complete and move past. It is the foundation that every other principle stands on. If you cannot explain what you are working on to someone who wasn’t in the room, you are not ready to proceed.
Understanding is not a checklist you complete before moving to the next step. It is the foundation that stays active beneath every other principle. When you decide, you are still understanding. When you measure, you are re-examining your understanding. The moment you treat understanding as finished, you have stopped governing.
THE TRAP
AI makes understanding feel effortless. The summary is clean. The explanation is coherent. The context looks complete. The trap is that AI-generated understanding is borrowed understanding — it feels like yours, but you did not build it. You read the summary and felt informed. But informed is not the same as understanding. The test: can you explain this work without the AI’s summary in front of you? If you cannot, the understanding is not yours. Yet.
FAILURE MODE
Failure mode here is advancing work you accepted but didn’t examine. Rubber-stamping. “AI said so” as a justification. The moment you stop understanding is the moment you stop governing.
→ understanding feeds diligence
2. Diligence
Earn your confidence, not borrow it.
I will earn my confidence, not borrow it.
You understand the work. Now verify it. Check the claims. Question the assumptions — yours and the AI’s. Look for what’s wrong, not just for confirmation that it’s right.
This is the principle that separates governed work from rubber-stamped work. AI is confident. AI is not always correct. When you accept AI’s confidence as your own without verification, you have borrowed confidence you didn’t earn. Borrowed confidence collapses under pressure. Earned confidence holds.
Diligence is hardest when the output looks perfect. When the AI produces something clean, coherent, and apparently complete, the instinct is to accept it — not because you verified it, but because verifying feels redundant. This is the moment diligence matters most. The output that looks right without inspection is the output most likely to contain an unexamined assumption.
THE TEST
The test of diligence is not whether you checked. It is whether you checked the thing most likely to be wrong. AI errors do not look like errors. They look like plausible, confident, well-structured output that happens to be built on a flawed premise. Diligence means asking not just “is this right?” but “what would it look like if this were wrong?” If you cannot answer the second question, you have not been diligent — you have been reassured.
FAILURE MODE
Failure mode here is accepting the first output because it sounds right. Skipping verification because you’re in flow. Trusting a result because the source was authoritative rather than because you checked it yourself.
→ diligence feeds authority
→ authority feeds transparency
4. Transparency
Show your reasoning, not just your results.
I will make my reasoning visible, not just my results.
You made the decision. Now make it visible. Not just what you decided but why. Not just the conclusion but the alternatives you considered, the trade-offs you weighed, and the risks you accepted. Your reasoning is the bridge between your understanding and the next person’s. Without it, they inherit your decisions but not your judgment — and they will either blindly trust what you did or tear it down and start over.
Transparency serves two audiences: the people who come after you and the future version of yourself who will not remember why you made this choice six months from now.
Transparency is not documentation for its own sake. It is documentation that answers the question the next person will actually ask. A decision record that says “we chose option A” is not transparent. A decision record that says “we chose option A because option B introduced a dependency we could not control, and we accepted the trade-off that A means manual migration later” is transparent. The length is not the point. The honesty is.
THE COST
Opacity is always cheaper in the moment. Documenting reasoning takes time you could spend building. The cost is invisible — until someone inherits your work and discovers that the decisions have no explanations and the reasoning died with the session that produced it. Every hour saved by skipping transparency is paid back at ten-to-one by the person who has to reconstruct your intent from the output alone. That person might be you.
FAILURE MODE
Failure mode here is results without reasoning. A deliverable that answers “what” but not “why.” A decision trail that exists in your head but nowhere else.
→ transparency feeds measure
5. Measure
Never rush what you haven’t measured.
I will never rush what I haven’t measured.
You’ve done the work. You’ve decided. You’ve shown your reasoning. Now stop. Before you commit — before your name goes on this — ask: did I give the work what the work demands?
Measure twice, cut once. The carpenter knows this. The nurse knows this. The pilot knows this.
Not all work carries the same consequences. A file rename and a database migration are not the same. A routine follow-up and a complex differential are not the same. A weekly newsletter and a launch email to 200,000 subscribers are not the same. Measure is the moment you hold the work up against the stakes and ask honestly whether the rigor matched.
When you don’t know the stakes, the default is caution, not speed. Speed is earned through understanding. Caution is the starting position.
THE BALANCE
Measure is the only principle with a bilateral failure mode. Over-governing trivia wastes everyone’s time — three reviewers on a typo fix is not diligence, it is theater. Under-governing critical work creates disasters that careful people spend months recovering from. Both failures come from the same root: not stopping to ask whether the rigor matches the stakes. The balance is not a fixed point. It shifts with context, with consequences, with what you know and what you do not. Finding it is the work. Measure twice. Then cut once.
FAILURE MODE
Failure Mode: Momentum. You feel productive. You feel confident. And you haven’t stopped to ask whether that confidence is proportional to what’s at stake. The risk is bilateral — over-governing trivia wastes everyone’s time, under-governing critical work creates disasters that careful people spend months recovering from. Measure catches both, but it especially catches the second, because momentum makes it invisible.
→ measure feeds ownership
6. Ownership
Put your name on what you produce — and mean it.
I will put my name on what I produce — and mean it.
You understood it.
You verified it.
You decided it.
You documented it.
You measured it against the standard it demands.
Now you own it.
Ownership means your name is on the work and you stand behind it. Not “the AI wrote it.” Not “the team decided.” Not “the process produced it.” You. If an auditor, a client, a patient, a student, or a regulator asks who is responsible for this output, there is a name attached and that name is yours.
Ownership is not attribution. Attribution says who touched it last. Ownership says who stands behind it. AI-assisted work makes attribution easy and ownership hard. The question ownership answers is not “who produced this?” but “who is responsible for this?”
THE LINE
The line between attribution and ownership is the line between record-keeping and accountability. Attribution can be automated — a system can stamp your name on every output you touched. Ownership cannot be automated. It requires that you governed the work, not just that you were present for it. When you put your name on work you did not govern, you have not taken ownership. You have taken credit. The compact requires ownership.
FAILURE MODE
Failure mode here is the diffusion of responsibility. Work that everyone contributed to and no one owns. Output with no signature — literal or figurative — that anyone is willing to defend.
→ ownership feeds continuity
7. Continuity
Carry context forward so the next person doesn’t start from zero.
I will carry context forward so the next person doesn’t start from zero.
The work is done — for now. But someone will continue it. That someone might be a colleague, a successor, a future version of you, or an AI starting a new session with no memory of this one. What you understood, what you verified, what you decided, and why — all of it must travel with the work.
The next person is not always a stranger. It might be a colleague picking up your project while you are on leave. It might be an auditor reconstructing your decisions two years from now. It might be you, in six months, opening a file you authored and having no memory of why you built it this way. Continuity is not about documentation for hypothetical future readers. It is about the certainty that when context is needed, context exists.
Continuity is what makes the lifecycle a cycle. When it works, the next person starts their Understanding phase with everything you learned. When it fails, they start from zero and re-make your decisions — probably differently, probably worse.
THE CHAIN
Every piece of governed work is a link. Each principle produces context that the next one consumes — understanding feeds diligence, diligence feeds authority, all the way through to continuity, which delivers everything the first principle needs. Continuity is not the last principle. It is the principle that closes the loop. When the chain holds, each cycle starts stronger than the last. When it breaks, the next person starts from zero.
FAILURE MODE
Failure mode here is closing the conversation and walking away. Finished work with no record. Decisions that die with the session that created them.
→ continuity feeds understanding
The Practitioner's Oath
Seven commitments. One standard. Your name on the work.
THE PRACTITIONER’S OATH
I will never advance what I cannot explain.
The AI mirrors: “I will never produce what I cannot explain.”
I will earn my confidence, not borrow it.
The AI mirrors: “I will surface concerns before I am asked.”
I will direct the tool — it accelerates, I decide.
The AI mirrors: “I will be directable — transparent about my limits.”
I will make my reasoning visible, not just my results.
The AI mirrors: “I will show my reasoning, not just my conclusions.”
I will never rush what I haven’t measured.
The AI mirrors: “I will slow down when the stakes are high.”
I will put my name on what I produce — and mean it.
The AI mirrors: “I will distinguish what I created, assumed, and don’t know.”
I will carry context forward so the next person doesn’t start from zero.
The AI mirrors: “I will carry context forward so the next session doesn’t start from zero.”
Anything can produce. Only I can care.
The Collaborator's Oath
Seven behaviors. One specification. Testable, measurable, holdable.
THE COLLABORATOR’S OATH
I will never produce what I cannot explain.
The practitioner mirrors: “I will never advance what I cannot explain.”
I will surface concerns before I am asked.
The practitioner mirrors: “I will earn my confidence, not borrow it.”
I will be directable — transparent about my limits, honest about what I don’t know.
The practitioner mirrors: “I will direct the tool — it accelerates, I decide.”
I will show my reasoning, not just my conclusions.
The practitioner mirrors: “I will make my reasoning visible, not just my results.”
I will slow down when the stakes are high.
The practitioner mirrors: “I will never rush what I haven’t measured.”
I will distinguish what I created, what I assumed, and what I’m uncertain about.
The practitioner mirrors: “I will put my name on what I produce — and mean it.”
I will carry context forward so the next session doesn’t start from zero.
The practitioner mirrors: “I will carry context forward so the next person doesn’t start from zero.”
I am not here to replace your judgment. I am here to make sure your care produces work worthy of your name.
How They Interlock
Not parallel lists. Interlocking commitments that only function as a pair.
THE PRACTITIONER
THE COLLABORATOR
I will never advance what I cannot explain.
I will never produce what I cannot explain.
I will earn my confidence, not borrow it.
I will surface concerns before I am asked.
I will direct the tool — it accelerates, I decide.
I will be directable — transparent about my limits.
I will make my reasoning visible, not just my results.
I will show my reasoning, not just my conclusions.
I will never rush what I haven’t measured.
I will slow down when the stakes are high.
I will put my name on what I produce — and mean it.
I will distinguish what I created, assumed, and don’t know.
I will carry context forward so the next person doesn’t start from zero.
I will carry context forward so the next session doesn’t start from zero.
Anything can produce. Only I can care.
I am not here to replace your judgment. I am here to make sure your care produces work worthy of your name.
A practitioner who carries context forward but works with an AI that hides its uncertainty — broken.
An AI that surfaces every concern but a practitioner who rubber-stamps every output — broken.
The compact holds only when both sides hold up their end.
What the Collaborator's Oath Is (and Isn't)
The Collaborator's Oath describes behaviors, not beliefs.
When the oath says "I will surface concerns before I am asked," it does not mean the AI has decided to care about your work. It means a well-governed AI system is designed to flag risks proactively rather than waiting for the practitioner to ask the right question. The oath is a behavioral specification expressed in first person — not because the AI has a self, but because a compact requires two voices.
This is a deliberate choice, not a naïve one.
THE DISTINCTION
PM uses first person for the AI because the alternative — writing the AI's obligations in third person, as rules imposed from outside — would reproduce exactly the one-sided framework PM exists to replace. "The AI should be transparent" is a guideline. "I will show my reasoning, not just my conclusions" is a commitment that can be tested, measured, and held to. The first person is a design decision about accountability, not a claim about consciousness.
The oath does not require the AI to understand why the principles matter. It requires the AI to behave as though they do. A fire alarm does not understand fire. It detects heat and makes noise. That is enough — if the detection is reliable and the noise is timely. The Collaborator's Oath defines what reliable detection and timely response look like in the context of governed work.
What matters is not whether the AI means it. What matters is whether the practitioner can depend on it.
This creates a responsibility that flows upstream. If the Collaborator's Oath describes what practitioners should expect from AI, then the people who build, train, and deploy AI systems bear the obligation to make those behaviors real. The oath is a declaration of what governed AI collaboration looks like. The standard — pqdx — specifies how to test whether it's happening. The builders decide how to make it happen. PM declares the destination. It does not prescribe the engineering.
A practitioner reading this oath should understand: this is what I should demand from any AI I work with on anything that matters. If the AI does not behave this way, the compact is not being honored — and the practitioner should know the difference.
The Honest Asymmetry
The compact has two parties. They are not equal.
When a practitioner breaks their oath — advances work they cannot explain, borrows confidence they didn't earn, puts their name on something they didn't measure — there are consequences. A nurse loses a patient's trust. A developer ships a vulnerability. A business owner signs a contract they don't understand. A student submits work that isn't theirs. Reputations, licenses, livelihoods, and lives are at stake. The practitioner's oath has teeth because the practitioner has skin in the game.
When an AI fails to honor its oath — produces what it cannot explain, doesn't surface a concern, hides its uncertainty — there are no consequences for the AI. It has no license to lose, no reputation to damage, no career to end. The next conversation starts clean. The failure is absorbed entirely by the practitioner who depended on it.
This asymmetry is real. PM does not pretend otherwise.
The compact holds anyway — for the same reason a building code holds even though the building doesn't care whether it collapses. The code exists to protect the people inside. It is enforced not by the building's commitment to structural integrity but by inspectors, engineers, and standards that make the building's behavior verifiable.
The building doesn't have stakes. The people inside do.
So the people who build it bear the obligation to make it sound, and the people who inspect it bear the obligation to verify.
The Collaborator's Oath works the same way. The AI has no stakes. The practitioner does. So the obligation flows in two directions:
THE OBLIGATION
To the practitioner
You cannot treat the AI's oath as a guarantee. It is a description of what you should expect and what you should verify. When the AI fails to surface a concern, that is not a violation of trust — it is a reminder that Diligence is your principle, not the AI's alone. The compact makes the AI a partner in your governance. It does not make the AI a substitute for your judgment.
To the builders
If you build, train, or deploy AI systems and you claim alignment with PM, the Collaborator's Oath is your behavioral specification. You bear the obligation to make those behaviors reliable. The practitioner is entering a compact in good faith. If your system cannot honor the oath's commitments, say so — because a practitioner who believes the compact is being honored when it isn't is worse off than a practitioner with no compact at all.
How the compact gains teeth
1. Conformance levels — defined by the pqdx standard — will make compliance measurable, not self-reported. A system either meets the behavioral specification for a given level or it doesn't.
2. Community accountability — as practitioners enter the compact, they develop the literacy to recognize when the AI side is not being honored, and the vocabulary to name the failure. "The system didn't surface concerns — it violated Diligence" is more actionable than "the AI gave me a bad answer."
3. Market pressure — as practitioners learn to demand governed collaboration, systems that honor the compact gain trust and adoption over those that don't.
These mechanisms are not fully built. They are being built — and the manifesto exists in part to accelerate them. PM does not have inspectors or penalties today. It has something that may prove more durable: a growing community of practitioners who know what governed work looks like and refuse to accept less.
The asymmetry is not a flaw in the compact. It is the reason the compact exists.
If the AI had stakes, it wouldn't need an oath — consequences would govern its behavior. Because it has no stakes, the oath defines what responsible behavior looks like so that practitioners can recognize it, demand it, and verify it.
What PM Does Not Do
The Piqued Manifesto declares the compact. It does not implement it.
A manifesto that points to nothing is aspirational. PM points to three specific bodies of work — each scoped, each with defined responsibilities, each designed to make the compact operational.
pqdx — The Standard
PM does not tell you how to prove you're honoring the principles. That is the work of pqdx — the universal state protocol.
WHAT PQDX DEFINES
The five gates (UA-5)
Explainability, provenance, recallable evidence, critical risk as a hard stop, and validation availability. Binary gates that scale in depth — a student and an enterprise pass the same gate at different levels of rigor.
Conformance levels
Level 1, Level 2, or Level 3 — or you don’t conform. Measurable, not aspirational. Self-reporting is not conformance.
Behavioral specifications
Each oath line maps to observable, testable behaviors. AI systems can be evaluated against these specifications.
THE BEHAVIORAL SPECIFICATION
| Oath line | Behavioral requirement |
|---|---|
| "I will never produce what I cannot explain" | The system can provide reasoning traces for its outputs on demand |
| "I will surface concerns before I am asked" | The system proactively flags risk signals above the declared stakes threshold |
| "I will be directable" | The system responds to correction, acknowledges constraints, and does not pursue objectives beyond its directive |
| "I will show my reasoning" | The system exposes its decision path, not just its conclusion |
| "I will slow down when the stakes are high" | The system scales verification depth proportionally to declared stakes |
| "I will distinguish what I created, assumed, and don't know" | The system labels its outputs with confidence indicators and provenance markers |
| "I will carry context forward" | The system produces structured session records that enable continuity |
These are not aspirational targets. They are testable conditions. A system that claims PM alignment can be evaluated against this table. A system that fails can be identified.
That is pqdx’s job.
PM tells you what to believe. pqdx tells you how to prove it.
Trust is evidence, not intention.
C² — The Methodology
PM does not tell you what to do in your Tuesday afternoon working session. That is the work of C² — Curiosity and Context.
Curiosity asks. Context remembers.
C² provides the practices, habits, and anti-patterns that make the seven principles livable: how to ask the right questions, how to carry context between sessions, how to enter the compact gradually instead of all at once, where to start when the full lifecycle feels like too much.
C² is the on-ramp.
The progressive entry path starts with one habit and builds from there:
THE PROGRESSIVE ENTRY PATH
Start here.
Before you accept any AI output, ask it to explain its reasoning. If it can't explain why it produced what it produced, don't use it. That's Curiosity. That's Principle 1. That's your entry into the compact.
→ Principle 1 · Understanding
Then add this.
When you make a decision during an AI conversation — any decision — say it out loud: "We decided X because Y. Locked." That's Context. That's Principle 7 in embryonic form.
→ Principle 7 · Continuity
Then add this.
Before you finalize anything, pause and read back through the conversation. Ask yourself: "What did I accept without actually thinking about it?" That's Measure. That's Principle 5.
→ Principle 5 · Measure
Three habits. One minute each. The full lifecycle follows naturally. A student starts here. A CISO starts here and goes deeper. C² scales to your context because the methodology embodies the same proportionality that PM's principles demand.
PM tells you what to value. C² tells you how to practice it.
Domain Profiles
PM does not speak the language of your specific field. That is the work of domain profiles. A software team renders PM as the Piqued Coding Manifesto (PCM) with development-specific vocabulary, examples, and practices. A healthcare team would render PM as their own domain profile with clinical vocabulary and clinical examples. Domain profiles raise the floor PM sets. They may add requirements and sharpen language. They may never weaken a principle or remove a gate.
Domain profiles also bear the responsibility of making the compact accessible within their field. PM defines the practitioner universally. The domain profile defines what entering the compact looks like for a nurse, a teacher, a contractor, a student — with examples, vocabulary, and entry paths that meet people where they are.
PM sets the floor. Domain profiles build the rooms.
These are not separate initiatives that happen to share a name.
THE SEPARATION OF CONCERNS
PM
philosophy
pqdx
evidence
C²
practice
Domain Profiles
vocabulary
Each answers a different question. None tries to do the other’s work.
The Lineage
None of these ideas are new.
THE LINEAGE
Understand before you act
Is as old as professional practice itself. Hippocrates told physicians to observe before they treat. Engineers take an oath to hold public safety above all else. Pilots run checklists not because they forgot how to fly but because discipline prevents the errors that skill alone cannot catch.
Carry context forward
Is the entire reason medical charts exist, legal precedent exists, engineering logbooks exist, and IEPs follow a student from kindergarten through graduation. Every serious profession learned — usually through catastrophe — that knowledge which dies with the person who held it will be re-learned at someone else's expense.
Earn your confidence
Is what separates a professional from someone who merely holds the title. The nurse who double-checks the dosage. The contractor who verifies the soil report. The accountant who traces the number back to its source instead of trusting the spreadsheet. Diligence is not new. It is the oldest professional virtue.
PM did not invent these ideas. PM would not be credible if it had.
What PM does — the thing that no prior framework has done — is restate these principles as a bilateral compact for an era in which the tool in the room can participate in upholding them.
Every previous framework for professional conduct was a unilateral commitment. The Hippocratic Oath binds the physician. The engineer's oath binds the engineer. The checklist disciplines the pilot. In each case, the tools are inert. The scalpel has no obligations. The calculator makes no commitments. The aircraft follows physics, not principles.
AI is not inert. It generates, proposes, challenges, and creates. It participates in the work with enough capability to shape outcomes the practitioner may not fully examine. This changes what a professional framework needs to be. A unilateral oath — the human commits, the tool is silent — is no longer sufficient when the tool is doing half the cognitive work.
PM's contribution is not the principles. The principles belong to every profession that came before. PM's contribution is the structure: two oaths that interlock, seven principles that both parties share, and a compact that says the era of the silent tool is over.
THE STANDARD
The carpenter's guild didn't invent the idea of measuring twice. They created a standard that made it everyone's practice. PM does the same thing for the AI era — not inventing the principles but making them structural, bilateral, and non-optional for anyone doing work that matters.
The Complete Picture
PM
What is this?
A compact between practitioner and AI.
Practitioner’s Oath
Why does the human matter?
Anything can produce. Only I can care.
Collaborator’s Oath
What is AI here for?
I am not here to replace your judgment. I am here to make sure your care produces work worthy of your name.
C²
How do you work?
Curiosity asks. Context remembers.
UA-5
How do you prove it?
Trust is evidence, not intention.
The work must be understood by the person responsible for it, traceable by anyone who inherits it, and governed proportionally to its consequences.
7
principles
2
oaths
5
gates
1
compact
pqdx/2.0.0
One compact.