Zero Neo

A business could always reproduce a great deal of what it buys. CRM, scheduling, forms, document management, collaboration, reporting, internal approvals, and most lightweight line-of-business work can be built out of capabilities already present in Microsoft 365 or Google Workspace. This was true before AI and almost nobody did it.

The reason was cost, not capability. Doing it required IT people, consultants, developers, administrators, integrations, and continuing maintenance. Specialized SaaS arrived preconfigured around a recognizable problem, so packaged configuration was cheaper than custom configuration. Buying was the rational choice, and for most buyers it still is.

Zero Neo asks whether AI changes that arithmetic. If a system can configure, connect, maintain, observe, test, and continuously improve capabilities a business already owns, then the default architectural question stops being which software to buy and becomes why new software is needed at all.

Zero Neo puts the options in order and requires each rung to be tried, or shown to fail, before the next one. The ladder is the whole doctrine in one picture.

Historically, someone proposing new software asks the room why they should not buy it, and the answer has to be argued by whoever is uncomfortable. That is the wrong direction. The proposal is what changes the environment, so the proposal carries the burden.

Zero Neo reverses the question to: why can we not accomplish the required business outcome with what we already have? New software has to earn its introduction. The answer may well be that it earns it easily, and that is a fine outcome for a review. What is not fine is never asking.

The hypothesis is easy to defend in a simple environment, so we test it in a hard one. Healthcare delivery organizations usually already own a productivity platform, an electronic health record or other system of record, identity infrastructure, email, calendars, forms, documents, collaboration, and communications. They then continue buying specialized products for patient engagement, scheduling, referrals, intake, care coordination, relationship management, messaging, portals, analytics, workflow, and now AI.

That makes it a good research case. It has complex workflows, privacy requirements, multiple roles, existing systems of record, human handoffs, and consequences when the work is done badly. If the decomposition holds there, it is worth taking seriously elsewhere. If it fails there, we want to know exactly which constraint broke it.

One workflow carries most of the argument. A referral runs from intake, through identity and eligibility, records retrieval, triage, scheduling, the visit itself, and completion with a report back to the referring clinician. Every step crosses an organizational boundary, and most organizations cannot say how many of their referrals close.

The failure is rarely a missing feature. It is fax, phone, portal, and inbox spread across systems that do not know about each other, with nobody able to see the loop end to end. Buying a referral product adds a system to that list. Zero Neo asks whether the loop can instead be made visible and accountable on the platform the organization already runs.

This is also where the cost of getting it wrong is documented rather than asserted. The note on what platform and SaaS choices cost care delivery collects the published evidence.

The methods in this program were not derived in the abstract. Labor as Code™, Operational Truth™, outcomes acceptance testing, and the worker contract were developed and tested in healthcare delivery work, where the workflows are long, the handoffs are between organizations, and a quiet failure has a patient at the end of it.

That provenance is a strength and a limit. It means the patterns survived contact with a hard environment. It also means they are shaped by that environment, and their transfer to other industries is a claim to be tested rather than assumed.

These are the open questions the program exists to answer. None of them are settled, and several may resolve against the hypothesis.

The obvious failure is that a badly assembled collection of platform components is cheaper on the invoice and far more expensive to operate. Licensing is the easiest number to compare and the least important one.

Zero Neo only holds if AI makes configuration, maintenance, usability, testing, and governance cheap enough to beat the packaged alternative on total cost. That is a falsifiable economic claim, and we intend to publish the cases where it fails as carefully as the cases where it works.

This is a research program, not a doctrine and not a product. Nothing here says specialized software is obsolete. It says new software is the last resort, and that the claim deserves testing rather than repetition.