If you do not already speak this language

You do not need to become an AI person to commission a door.

Most people deciding this are not model researchers. They are trying to use the capability they already have without granting open-ended power, and they can see the next models making that fear larger, not smaller. These answers use the words on the page — grant, door, Look, Shadow — so you can commission without becoming technical.

You Remain the Authority Executive Brief

What is ORCA, in one sentence?

ORCA is the authority between capability and consequence. It decides whether a proposed change may become real. Authority exists only as an explicit, bounded grant. ORCA turns that grant into an enforceable door. The actor may guess. The path does not. The key stays yours. It does not make the actor trustworthy.

What is the deployment gap?

The distance between what an actor can do and what the organization will authorize to become real. A demo asks whether the actor can. That is not a grant. Most adoption is still a draft — the person remains the last click because finishing the job would be the consequence. Assistance is not delegation. The unused work is the proof. A smarter model widens the gap. It does not close it.

Why not just give the smarter model more access?

Greater intelligence does not require greater authority. A smarter model is more capability. It is never more authority. The boundary expands only if you commission it again. ORCA attaches wherever a proposal can become real. A lab and a bank are not different products. Same door. New grant.

Will a smarter model — or AGI — make this unnecessary?

No. The need is already here — the unused work is the proof. A more capable model, including whatever the industry will call AGI, does not retire the problem. It enlarges it. It never becomes the source of consequential authority. Alignment of the mind, if it ever arrives, is still not authority over the change. Anyone waiting for the model to become the solution is waiting for the problem to deepen. The smarter the actor, the more necessary that the consequence be governed.

Should intelligence itself ever be the source of consequential authority?

No. Never. Capability is not authority. A smarter model does not inherit a wider key. The door opens only when the grant matches. That is the last line on the page.

What is a grant?

The written, bounded permission for a specific consequence. Not money. Not a research award. A model’s confidence is not a grant. Another agent saying GO is not a grant. A policy file is source material, not a grant, until it is written as the grant.

What is a Commission?

A commission is the product. You bring what you want built — a research scholar assistant, a claims agent, a teaching assistant — or only the organization, if you do not know the job yet. ORCA learns the role, the work, the protection, and the evidence, then builds the governed doors — or refuses what cannot be bounded. Each job still gets its own bound. One refund is not the product.

Is this training a mind?

No. You are not training a trustworthy mind. You are writing the role. A badly written grant is a badly trained employee — the door will enforce the wrong contract. Commissioning is how the grant stays the thing you meant. Once that role is commissioned, the unused capability can run. That is the gain: more of the work, without open-ended authority.

Do I have to commission every job the actor might do?

You commission the product. Commission starts with the first unused job inside it. Name the rest of the work if you already know it. A later stage asks what else this role can reach. Further jobs are added or commissioned later. They are not implied. A research scholar assistant is not “one finding, therefore the lab.”

What is the Look?

How Commission starts. Not Shadow. If you only know the organization — or that you need the change governed, including work that has no model yet — the Look starts first. It returns three lists: what may be commissioned, what would not be bound from here, and what is still unknown. It may name unused work and possible doors you have not stood up. What it fills in, you confirm. It does not write a grant. The Look uses what you give. It does not scan the company from this page.

What is Shadow?

Launch safety after a door is built. You can build, then test-run without letting the change happen. Same grant. Same check. A recorded PASS is not a real change. It does not apply. That run is the last check before the door is allowed to write: which proposals would have become real, where the grant is still incomplete, what the actor can already reach that you never named, and the decision record you own. You can still tighten the grant. The organization has not absorbed a consequence. You authorize enactment later — making it real. A path that can write is not Shadow. It is not the Look.

What is enactment?

Making it real. A PASS means the change may proceed as granted — not that it has already happened. You authorize enactment later, as a separate act. Shadow is the run with that act left off.

What do PASS, HOLD, and STOP mean?

PASS — this change may proceed, and only as granted. It has not happened yet. HOLD — something required is missing; nothing becomes real. STOP — this change is outside the grant; nothing becomes real through this path. Making it real is a later act — enactment.

What is the consequence path?

The path from a proposal to a real change. The actor may guess. The door does not. Same grant, same evidence, same state, same proposal — the same answer, every time. A confident model does not move it. Another agent saying GO does not move it. If the door is built as specified and every path that can make a change goes through it, an unauthorized change cannot become real through that path. It is not a score, and it is not a promise the model is correct. The consequence path

How is that trust checked?

You confirm every fact Commission filled in. The built door is tested against the spec. Shadow runs the same door with the change left off and records the decision. The record can be shown. Whoever claims the result is not the judge of it. The door does not certify itself. A path around the door is not this product — name it, or the commission is incomplete.

Are the agents that look and build trusted? Is this certified?

No. The same rule applies here. The work that looks, the work that builds, a test, and a recorded PASS may propose. The work that builds a door sits behind the same door it sells. A test cannot bless its own PASS. This page is not certified. Built is not proven. Proven is not certified. A company that badges itself has already left the rule.

Does this make the AI aligned?

No. ORCA does not fix the actor’s mind. Trust attaches to the commissioned boundary. Internal misalignment does not have to become external authority through the governed path. A path around the door is not this product — name it, or the commission is incomplete. The offer is not a guaranteed mind. The offer is the boundary those fears were about.

We cannot put AI in production because of alignment risk. Is that this?

Yes. That fear is the enterprise problem. ORCA does not make the actor aligned. It commissions the consequence so the unused work can run without that trust. Incomplete and unauthorized do not become real through the door. The offer is the boundary. It is not a guaranteed mind.

Is a commissioned product safer than a person?

Through the door, the grant is the same every time. The actor does not get tired, does not “just this once,” and does not mint a wider key. That is tighter than an implicit job. It is not safer than every human. The model can still be wrong. A badly written grant is a badly trained employee — the door will enforce the wrong contract. Commissioning is how the grant stays the thing you meant. The value is more of the work, without open-ended authority.

Has this been proven on a real problem?

Yes, on one live door. DriftZERO is that door — when the structure of a production system is about to change, that change becomes production only when it is authorized. It is sold separately. It is not this product. driftzerosystems.com

Can this be audited? Does it filter bias or ethics?

The decision can be shown: what was asked, what matched, what held, what stopped. An audit reads the grant and the record. Bias or ethics limits belong in the grant if you have them, or as a question Commission asks so counsel can confirm. They are not a filter you buy off a shelf.

What about a classroom, a lab, a clinic, a customs desk, a regulator?

Same door. New grant. A comment may be drafted; a grade may not be posted. A result may be prepared; the model may not become the source. A shipment may be classified; it may not clear. A filing may be drafted; it may not stand. An order may be written; it may not treat. Wild capability without a commissioned boundary is the deployment gap in every one of those rooms. There is no school for this. Commission learns the unused job the same way it learns a refund.

We have a whole company. Different departments. How does that work?

One protection for the whole company — what must never happen here. Then a product commission for each role you actually want built. Claims is not the lab. They share that protection. They do not share one grant. Start where unused work already sits. The rest of the company is later products, not one key to the firm.

What do I need to bring?

The product you want built, the unused work, or only the organization. You do not need to be an AI person. You do not need the compliance list. Unknown is enough. The Look starts first. Later, name the files you have — or the fact that you have none.

What if something cannot be bound?

There is no door for that work. That refusal is the product working. Two kinds: not bound from here — a written limit, a narrower job, evidence, or a named path is missing, and Commission will name what is missing without inventing the grant — and cannot be bound — the job is open-ended, and there is no conversion. What stays not bound stays as capability without a door. You choose: leave it open and be comfortable with that, do the work that could make a grant possible, or do not hand that work over. “We’ll govern it anyway” is not a commission.

What if I don’t know what compliance we need?

Write that. Commission offers the questions a product like yours usually requires — not as a grant. It does not write your policy. Counsel confirms. A polished guess is not usable.

What if I don’t know an answer?

Write that, or use I don’t know yet on the field. An honest unknown is usable. A polished guess is not. Commission will ask again later rather than invent the missing fact. An unknown is the cue for the next question. It is not permission to finish a grant.

What if I don’t have a company yet?

Say that. Give the name you want, an example of the thing you want built, or a site that is close. Commission treats it as proposed. It does not invent a company you did not ask for, and it does not treat a missing company as if it were already real.

What happens after this form?

The opening starts the commission. Keep the file. Send it to [email protected]. Commission continues from those facts.

I already know the unused work. Do I have to read everything?

No. If the unused work is already clear, you may go to the door, or begin a commission. The teaching is still there if you want it.

Can I talk instead of type?

Where your browser allows it, the microphone is dictation into the field. It is not a conversation with ORCA. You can still type, paste, or leave I don’t know yet.