Intent to production

The route begins with someone who wants a need solved and looks for a service that can take care of producing the software. They can buy an advance that makes the solution understandable first, learn the conditions for continuing, and choose what to authorize. This page connects that experience to the mechanisms of the model. It uses the shared case described in the reference model overview.

A simple intention, an organized service

A barbershop wants to let customers book appointments online. Its intention is to stop coordinating times by hand; it has not decided every behaviour and has not bought the whole construction. The demand is given an identity, Delivery Unit DU-001.

The provider offers a bounded discovery: clarify the rules, produce a prototype, examine the architecture and present conditions for continuing. The barbershop can buy this advance without choosing who performs each activity. The interaction can be an assisted conversation or a self-service interface; the model does not depend on a portal.

Offering this first advance means producing learning and a usable decision at a cost that suits a small demand. The enabling hypothesis is that AI applied to the questions, the alternatives, the checks and the preparation of context reduces that effort. The commercial relationship in the case follows from that change in production: the provider can consider a bounded entry point without having to recover all of the preparation through construction that has already been promised.

What is bought before building

Commitment C0, at a fictional price of 20, covers clarification S1 and design S2. Its outputs are the rules, a prototype, a technical design, the dependencies, and enough information to discuss the next offer. The price pays for this advance even if nothing is built afterwards.

Hoping to reach usable software one day does not widen the commitment. The barbershop can receive the accepted outputs, decide not to buy another step, and close a fulfilled discovery. That is different from interrupting construction that has already been agreed, before its conditions are met.

The rights to use the outputs, and the conditions of their delivery, should be clear in the agreement. Here the barbershop receives the references and artifacts from discovery to guide its decision and any continuation, including with another provider under the agreed conditions. The usefulness of those outputs never becomes an obligation to buy the construction.

Clarification and business decisions

S1 turns the intention into usable rules. The case assumes a 30-minute service, a chosen professional, and working hours registered in advance. Intervals for the same professional may not overlap. A confirmed cancellation frees the interval, and different professionals may work at the same time.

The barbershop makes, or delegates, the business decisions this requires. In a hybrid configuration it takes part in the clarifications it has kept for itself. In an autonomous configuration a delegated authority resolves the choices covered by prior criteria. A question outside that scope goes to the appropriate authority, which may itself be another system with a sufficient mandate.

Output A1 gathers the rules and their limits. Its acceptance, Q1, is a Quality Gate: it recognizes that the clarification is fit for its purpose, on an identifiable basis and by an identifiable authority. Producing a text is not enough to accept a rule that still contradicts itself.

Prototype and architecture produce information together

S2 combines prototype and architecture. They can proceed in parallel, provided they share the same rules and reconcile their differences. A choice in the interface can reveal a technical dependency; a constraint in an integration can force the proposed interaction to change.

The integration identified here is the barbershop’s calendar service, which stores bookings and cancellations. The design has to explain how two simultaneous confirmations of the same interval will be prevented. The prototype makes it possible to discuss selection, confirmation, unavailability and cancellation before the full solution is built.

A2 brings these outputs together with the approach to evaluation, and Q2 accepts the design for its purpose. Progressive Estimation uses the information to revise forecasts, and Pricing turns the conditions of an offer into a price and a set of responsibilities. What was learned may reveal more work, or that the idea is not viable. It does not produce automatic certainty about future cost.

The offer and the choice to continue

The provider offers booking, implemented, validated and deployed, for 80. Reminders would add 15, making an alternative at 95. Advance payment for the barbershop’s services is left out of both offers. All values are fictional and bear no relation to market prices.

The Commercial Gate lets the barbershop authorize booking, ask for an adjustment, defer, or not continue. A partial approval means selecting an option that can actually be delivered: booking works without reminders, while reminders depend on existing bookings and are not sold here as a solution on their own.

On the main path, the barbershop chooses the offer at 80. That creates C1, covering implementation S3, integrated validation S4 and deployment S5. Q2 had not authorized any of these. Asking for an adjustment would not have counted as accepting an offer that was still under revision.

The same choice can be delegated

The barbershop can grant a mandate to commit up to 90 more after discovery, as long as specified technical and commercial conditions hold. The authorized system chooses booking at 80; the alternative with reminders at 95 exceeds the mandate. Ten remains as authority that has not been committed.

The aggregate contracted price is 100, and the aggregate authorized ceiling is 110. Neither figure is the provider’s internal cost or a payment already made. Billing and payment follow the agreement, separately from the events of authorization and acceptance.

A fully autonomous path does not concentrate every authority in whoever executes. The buyer, the provider and the environment can each delegate different decisions. The buyer’s economic authorization does not allow the provider to issue arbitrary offers, nor to act in production without the corresponding authority.

Production and evaluation under the chosen commitment

The Execution Policy may combine professionals and agents, or allow production and decisions to run autonomously. The Orchestrator keeps the context, checks conditions and coordinates continuity; Routing mobilizes eligible Execution Resources, dedicated or organized in Execution Pools.

S3 produces A3, an identified implementation. Q3 recognizes the local checks that were planned, without assuming that integration has been shown to work. Here S4 waits for access to the calendar’s validation environment. Having budget or available executors does not remove that dependency.

Once access is granted, evaluation finds two confirmed bookings for the same professional in overlapping intervals. The divergence breaks A1, so the flow returns to S3 for a correction covered by C1. The new revision goes through the relevant checks and returns to S4. Q4 is granted only when the evidence supports the integrated criteria.

Deployment and the end of the commitment

S5 uses the accepted revision and the environment’s operational permission. That permission can be given in advance through a delegated policy or by a direct decision. A human intervention at this point is not assumed.

A5 records the deployment, the checks and the continuity information that was agreed. If the action loses confirmation, its effects have to be established before it is repeated. Q5 accepts the stage within the contracted scope, and closing the commitment also checks remaining obligations.

The result is booking that can be used under the defined conditions. Less manual work and more revenue remain benefits to be observed, not consequences proven by acceptance. Ongoing monitoring and support are part of the service only when there is an explicit commitment to them; checking the deployment does not create unlimited operation.

A capacity that can serve again

While DU-001 moves forward, the provider can carry other demands. Persistent context, the distribution of work and reusable criteria make it possible to examine a recurring production capacity rather than the organization of an isolated project. Sharing resources does not authorize sharing data or decisions between clients.

The hypothesis enabled by AI is that interpreting, experimenting, producing, evaluating and coordinating routes like this one would need less continuous human direction across a growing range of services. The architecture admits hybrid and autonomous operation, but the example does not show lower cost or real scale. That difference would have to be measured with the attempts, the waiting and the discoveries that ended without construction all included.

A demand in industry could involve more authorities, integrations and constraints. It would keep the same relation between purchased advance, information, offer and decision, with a different cost and policy. The simple case does not limit the model to small businesses, nor does it assume that every complexity can be absorbed economically.

Boundaries

This route is proposed and entirely fictional. It describes one possible path; production is not the destination of every commitment. The identifiers organize the reference without creating new terms or a universal sequence. The reference pipeline details its five stages, and the example Delivery Unit gathers the same case with its amounts, evidence and alternative endings.