What is it for?
Stated as an outcome, not a category name.
Operating Theory · September 2026
The business application polypharmacy problem
Applications accumulate the way medications do. Each addition was reasonable at the time; the collection is nobody's deliberate design and nobody's responsibility to review.
Under the pillar Zero Neo
Accumulation
A twenty-person company can easily be running separate systems for CRM, customer communication, email marketing, forms, surveys, document management, electronic signatures, scheduling, booking, task management, project management, knowledge management, customer portals, workflow automation, analytics, dashboards, employee onboarding, customer onboarding, support tickets, relationship management, internal approvals, notifications, lightweight databases, content publishing, and now AI assistants.
Each purchase probably made sense on the day it was made. Somebody had a problem, found a tool, and solved it. No single decision was wrong.
The collection is a different matter: subscription expense, duplicate data, multiple identities, fragmented workflows, integration requirements, security surfaces, training burdens, vendor dependencies, inconsistent interfaces, abandoned features, conflicting records, and operational ambiguity about which system is telling the truth.
Reconciliation, borrowed from medicine
Clinicians do not assume that every medication a patient has accumulated over fifteen years still belongs in the regimen. They reconcile: list everything, ask what each item is for, check for interactions, and stop what is no longer earning its place.
Businesses should do the same with applications, on a schedule, with the same posture: not hostile, just unwilling to assume.
What outcome does it produce?
Something the business would notice losing.
Who actually uses it?
Measured from sign-in and activity, not from the licence count.
What capabilities does it provide?
Decomposed into forms, records, workflow, messaging, storage, reporting.
Do we own those capabilities already?
Somewhere else in the platform, or in a system of record.
Could AI configure them instead?
At what build cost, and at what ongoing maintenance cost.
What breaks if we remove it?
The honest answer is sometimes nothing, and sometimes a great deal.
The argument, stated carefully
The claim is not that applications should disappear. Plenty of them are the best available answer and will stay that way.
The claim is that every application should keep earning its place, and that almost no organization has a process which asks it to.