Bounded service · Candidate · unavailable

Pilot Decision Protocol: Design a bounded test that can support stopping, revising, or further investigation.

For: A named owner considering one bounded intervention.

Choose this when

A testable question may be sufficiently defined for protocol design.

What happens next: Proceed only when evidence and operating conditions support another bounded step.

Accepted scope

What enters the engagement.

One bounded intervention
Measures and stopping conditions
Affected-party and non-AI considerations
Fallback and review ownership

How the work moves

Define the decision question
Design the protocol and safeguards
Collect agreed evidence
Make an accountable stop, revise, or further-test decision

Do this today

Something that already does most of this, free.

A candidate package you cannot buy is worth very little on its own. This is the part of the same job that is finished, open, and available in a browser right now — no account, no email, nothing transmitted.

Available now

Precommit the pilot with the Pilot Design Kit

Bounded purpose and hypothesis, baseline and measures, affected-party input, stop conditions, rollback and fallback, and an end-of-pilot decision firewall that exists specifically to stop a pilot from becoming production by default. Written before the test, exported as a file, and not editable into a different story afterwards without the change being visible.

Open the Pilot Design Kit →

What the paid version would add

The part a browser tool cannot do.

Someone independent of the pilot's sponsor holding the stop condition. The kit can write one down; it cannot make anybody honour it when the pilot is going badly and the person who championed it is the person deciding whether it stopped. That is the whole difference, and it is a governance arrangement rather than a document.

Why it is not bookable

The specific things that are not decided.

“Candidate format” is usually a way of saying “coming soon” without a date. Here it means the list below: each item is genuinely open, and publishing a price before settling them would be selling a shape rather than a service.

Whether the pilot itself is in scope. Designing a protocol and running one are different engagements with different risk, and combining them puts the designer in the position the stop condition exists to guard against.
Open
Who holds stop authority. If it stays inside the sponsoring team, the arrangement adds nothing. If it moves outside, that has to be agreed by someone senior enough to make it real.
Open
Measurement support. Whether baselines are collected, verified, or simply accepted as stated. Accepting a stated baseline makes the whole result unfalsifiable.
Open
Duration. A protocol takes days; a pilot takes a quarter. Publishing one price for both would be a fiction.
Open
That a pilot is not a quiet path to production. Settled, stated in the free kit, and not negotiable in any paid version.
Settled

What the buyer receives

Named artifacts—not vague transformation.

Protocol record

Measures and limits

Evidence record

Decision and review triggers

This named package is not bookable yet.
Its package-specific inquiry or qualification process is inactive. You may still bring RN the underlying problem for a separate, nonconfidential fit conversation. Any resulting work would receive its own written scope, deliverables, boundaries, timing, and price before it begins.

Aloha AI provides consulting and education, not legal representation, professional certification, or guaranteed outcomes.

Bring RN the underlying problem Compare all ways to work together