The Maintenance Gap
Personal AI capability will not be distributed fairly if its care work is treated as a private hobby instead of a condition of meaningful participation.
01 / the concealed labor
Access is the beginning, not the benefit.
The maintenance gap is the distance between being allowed to use an AI capability and having the time, support, authority, and continuity to keep it useful.
That gap is easy to miss because a capable application can make the first encounter feel complete. Someone receives an account, learns a few commands, and produces an impressive result. But a stack is not a one-time result. It is an evolving arrangement of context, habits, sources, permissions, judgment, and tools. Each part ages at a different speed.
Stack People are the people who can shape that arrangement over time. They are not simply App People who have learned more features. App People adapt their work to a product’s fixed workflow; Stack People assemble methods that can be revised when the workflow stops serving the work.
The central inequality of the app-to-stack transition may be maintenance: some people will inherit time, coaching, trustworthy context, and the right to improve a capability, while others receive only the obligation to use one.
This is an argument about a likely structure, not a claim that every organization is already there. The direction matters because many institutions are beginning to hand out access before they have decided who carries the repair work.
02 / allocate the care
A capability has more than one caretaker.
Use the ledger to bring one responsibility forward. The cards do not compete for ownership; they make a shared system legible. A personal stack becomes fragile when every kind of care quietly lands on the person with the least spare time.
Entry 01 / person
Keep the question yours.
The person remains responsible for the aim of the work: what they are trying to decide, whom they owe care, what cannot be disclosed, and when an answer should be rejected. That is not busywork. It is the part that keeps a stack from becoming somebody else’s default.
Personal care also includes portable habits: recording a useful prompt, naming an uncertainty, and preserving the reason a method exists. Those habits can travel without carrying a client file, an employer’s data, or a promise made in the wrong setting.
Entry 02 / employer
Fund the upkeep you depend on.
If an organization expects people to produce work through AI-enabled methods, it cannot call the upkeep extracurricular. Time to test changes, clarify policy, maintain approved context, and learn from a failure is part of the operating cost of the capability.
Employers also need to distinguish a person’s portable practice from organization-governed material. A useful stack should help someone work well inside a role without quietly requiring them to surrender their accumulated judgment when the role ends.
Entry 03 / institution
Make repair a public skill.
Schools, professional bodies, libraries, and workforce programs can teach more than tool operation. They can teach how to inspect a result, keep sources visible, preserve a handoff, and change a method without erasing the human question.
That support will matter most where people do not have paid experimentation time, a technical peer group, or a private archive of reliable context. A society that treats maintenance as invisible will reproduce advantage under the friendlier name of access.
03 / a maintenance practice
Build care into the work order.
A practical response is not to turn every person into a systems administrator. It is to make the upkeep small, named, and shared. A team can set a regular moment to review where a method learned from, what it now assumes, and which human can alter its next step.
This is where the app-to-stack transition becomes organizational rather than individual. A stack does compound capacity, as The First Stack Generation argues, but compounding is not automatic. What compounds can be good judgment and reusable practice, or unexamined error and dependence. The difference is maintenance.
04 / the real promise
Capability should not require private abundance.
The better promise of Stack People is not that everyone becomes a lone operator. It is that more people can retain a growing way of working without being trapped by the application that first enabled it.
That promise needs supports: protected time, legible rules, portable records, trusted places to practice, and an honest division between what a person owns as method and what a relationship governs as context. None of these are glamorous. All of them decide whether a capability is durable.
The first stack generation will be judged partly by what it makes possible and partly by whom it asks to maintain the possibility. If the repair work stays hidden, the new stack will recreate an old bargain: powerful systems for those with slack, fragile dependence for everyone else.
App to Stack / closing note
Do not confuse a login with a livelihood.
A login opens an application. A maintained stack gives a person a continuing capacity to understand, alter, and carry their work. The work of making that capacity durable is where the next social contract around AI will be written.