The person becomes the platform
The First Stack Generation
How the Missing First Rung Creates the New Digital Class Divide by 2031
Person as platform
The labor market is not failing equally.
Most established workers remain employed. The national unemployment rate remains relatively low. Businesses continue investing, consumers continue spending, and the American economy continues growing.
But the doorway through which young people traditionally entered professional life is narrowing.
In the first quarter of 2026, unemployment among recent college graduates—those between 22 and 27—was approximately 5.7%. Among those who were employed, 41.5% were underemployed, meaning they worked in occupations that typically did not require a college degree.1
RBC identifies the same structural tension. White-collar employment is declining, hiring is concentrated in relatively few sectors, and AI-related infrastructure is driving nearly the entire business-investment story while producing comparatively few traditional entry-level jobs. RBC describes this as a significant skills-matching problem for new graduates.2
The usual interpretation is that young people are temporarily having difficulty finding good jobs.
I believe something larger is happening.
The disappearance of the entry-level job is colliding with the transition from apps to personal stacks. Together, these forces will create the first generation whose economic identity is not defined primarily by one employer, one job description, or one collection of applications.
By 2031, the most important divide among young workers will not be between people who use artificial intelligence and people who do not.
Nearly everyone will use AI.
The divide will be between App People and Stack People.
The App Person enters an institution’s system.
The Stack Person arrives with a system.
The Old Agreement
The traditional employment relationship had two parties:
- the person;
- the institution.
The institution owned the building, equipment, information, operating procedures, customer relationships, and software. The person entered that environment and learned how to function inside it.
The company was the highest-level organization.
The employee learned Microsoft Office, Salesforce, Workday, Slack, Adobe, QuickBooks, SAP, or whichever applications the company had selected. The employee’s work was divided among these applications, and the employee’s professional memory became scattered across documents, inboxes, folders, dashboards, and databases owned by the employer.
When the employee left, most of the accumulated digital capability remained behind.
The employee carried away general experience, but not the operating environment in which that experience had become productive.
The next job began with another login screen.
New applications. New folders. New procedures. New data structures. New institutional memory.
The worker started over.
That was the App Person.
The App Person could become highly skilled, but the skill remained partially dependent on access to an institution’s systems. The company was the platform. The employee was a user.
This arrangement worked particularly well for young graduates because the company also provided the first rung of the career ladder.
The graduate arrived with a degree but little practical experience. The company supplied the environment, assigned relatively simple work, supervised the result, and gradually developed the person into a professional.
The young employee gathered information, prepared presentations, updated spreadsheets, wrote first drafts, summarized meetings, performed routine analysis, and supported more experienced workers.
The work created value for the company.
The repetition created expertise for the employee.
Artificial intelligence now threatens that arrangement from both directions.
It absorbs much of the routine work that justified hiring the beginner, while remote work makes the human apprenticeship surrounding that work harder to provide. New York Fed researchers estimate that remote work could explain 64% of the recent increase in unemployment among young college graduates because managers find it more difficult to train and mentor inexperienced employees in distributed environments. Their analysis suggests that remote work, rather than generative AI, caused most of the initial deterioration.3
AI did not create the missing first rung.
AI arrives just as that rung is already weakening.
The Missing First Rung
The old entry-level bargain was straightforward:
Perform the organization’s lower-level work, and the organization will teach you how to perform higher-level work.
But the lower-level work is precisely where the synthetic subconscious is strongest.
AI can prepare the first draft.
It can organize the spreadsheet.
It can summarize the meeting.
It can research the competitor.
It can update the report.
It can produce the presentation.
It can answer the routine customer question.
It can convert scattered information into a coherent briefing.
The experienced employee still provides judgment, verification, context, responsibility, and relationships. But that employee may no longer require three junior employees to prepare the underlying materials.
The employer therefore faces a new calculation.
Should the company hire an inexperienced graduate, train that person, supervise the work, provide benefits, and wait for productivity?
Or should it give the existing experienced employee a capable synthetic subconscious?
Increasingly, the company will try the second option first.
Early job-posting evidence does not yet show a distinct economy-wide collapse in AI-exposed occupations. Overall hiring has slowed more broadly, and researchers caution against blaming AI for every weakness in the labor market.4
But AI does not need to eliminate millions of jobs immediately to change the entrance into professional life.
It needs only to alter the decision at the margin.
One department delays one graduate hire.
Another replaces two vacant positions with one experienced employee and an AI system.
Another eliminates its formal training program.
Another decides it needs contractors rather than a full-time junior analyst.
The change accumulates quietly.
There is no dramatic announcement that the first rung has been removed.
There are simply fewer places to put your foot.
From Apps to Stack
This is where the transition from apps to stack becomes economically important.
A personal stack is not a larger collection of apps.
It is the opposite.
The App Person has a collection of vendor-defined systems. Each application owns a fragment of the person’s digital life. One application contains email. Another contains documents. Another contains financial records. Another contains projects. Another contains contacts. Another contains creative work.
The person moves among the applications, repeatedly translating context from one system into another.
The applications remain permanent.
The work passing through them is temporary.
A personal stack reverses that architecture.
The person becomes the highest-level organization.
The stack contains the person’s enduring data environment, memory, methods, procedures, permissions, standards, relationships, project histories, and synthetic subconscious. The model predicts, but the stack remembers.
Software becomes temporary.
When the person needs an interface, the synthetic subconscious creates or assembles one. When the task is complete, the interface can disappear. The information, procedures, and accumulated capability remain in the person’s stack.
The App Person opens a tax application.
The Stack Person says, “Do my taxes.”
The synthetic subconscious gathers the relevant records, applies the appropriate procedures, creates whatever temporary software is necessary, identifies uncertainty, and presents the decisions requiring human attention.
The tax app is no longer the permanent center of the activity.
The person is.
This architectural shift changes what a young worker can bring into the labor market.
The App Person brings familiarity with tools.
The Stack Person brings retained capability.
The App Person says:
I know how to use the applications your company has selected.
The Stack Person says:
I operate a system that can connect to your company, learn its required form, maintain an outcome, and improve through experience.
The App Person applies for a position.
The Stack Person connects a capability to an economic need.
Underemployment Becomes the Incubator
Underemployment is normally treated as a failure to match education with work.
A graduate studied finance but works in hospitality.
A communications graduate performs administrative work.
A computer-science graduate handles technical support.
A business graduate becomes a retail supervisor.
The graduate is employed and being paid, but the job does not use the person’s full educational or productive capacity.
That is a real problem.
But the personal stack creates another way to interpret it.
Underemployment means there is unused capacity.
The organization is paying the person, but the job is not consuming everything the person can do.
In the App Era, much of that unused capability remained trapped. The employee still had to sit inside the organization’s applications, remain available, repeat administrative procedures, and wait for the institution to assign something more meaningful.
In the Stack Era, the synthetic subconscious begins absorbing the predictable portions of the job.
The employee still performs the work.
The employee still accepts responsibility.
The employee still protects confidential information and keeps every promise.
But the amount of conscious attention required to maintain the outcome falls.
The underemployed worker gains excess productive capacity.
That capacity can be wasted.
It can be consumed by additional meetings, constant inbox checking, artificial busyness, and work expanded merely to fill the available day.
Or it can be directed toward another legitimate economic relationship.
The communications assistant begins maintaining a newsletter for another organization.
The technical-support worker manages AI workflows for two small businesses.
The retail supervisor produces weekly inventory analysis for several stores.
The finance graduate maintains reporting systems for local entrepreneurs.
The young worker does not necessarily leave the original job.
The job becomes the first operating environment in which the stack begins learning.
Underemployment becomes the incubator for overemployment.
Not the dishonest version of overemployment in which a person secretly accepts several incompatible full-time jobs and makes promises that cannot all be kept.
The legitimate version is a portfolio of clearly bounded responsibilities.
One employer.
Two clients.
A recurring research product.
A small digital service.
A temporary project.
Several sources of value supported by one compounding personal stack.
The worker is not selling the same hour several times.
The worker is reusing capability.
The Timeline
The following is a prediction, not a claim of certainty.
Assume that the number of serious personal-stack users grows rapidly and reaches approximately 300 million worldwide by 2031. The transition will not occur evenly, and traditional apps will remain profitable and widely used. The App Economy will fight back with better interfaces, more AI features, stronger ecosystems, and deeper enterprise integration.
Apps will not disappear.
But the role of the app will change.
2026: The Doorway Narrows
In 2026, the deterioration remains easiest to see among young graduates.
Companies continue employing their existing workers but become cautious about adding inexperienced people. Entry-level hiring weakens without producing a corresponding surge in broad unemployment.
The labor market feels stable to the person already inside and hostile to the person trying to enter.
Graduates respond by sending more applications.
AI helps them produce polished résumés, customized cover letters, interview answers, and follow-up messages. Because the same capability is available to nearly everyone, application volume rises while differentiation declines.
Employers receive more professional-looking applications containing less useful signal.
The résumé becomes inflated.
The first young Stack People begin responding differently.
Instead of using AI only to describe themselves, they use it to build something that already works.
They publish recurring research.
They operate a small customer-response system.
They maintain a real website.
They develop an industry knowledge base.
They produce market intelligence.
They serve one small client.
They create evidence that their synthetic subconscious can maintain an outcome over time.
The first divide appears:
AI-assisted applicants versus AI-augmented operators.
2027: Proof Replaces Promise
By 2027, employers become less impressed by AI-polished applications.
The cover letter becomes almost meaningless because every candidate can produce a competent one instantly. Résumés remain necessary, but employers assume that language has been optimized.
Hiring begins moving toward proof.
Candidates are asked to complete paid trials, operate simulations, demonstrate portfolios, diagnose live problems, or show systems they have maintained.
The strongest young candidates do not merely show finished artifacts.
They show continuity.
They can demonstrate that their system produced a weekly briefing for six months, maintained a customer workflow, tracked a changing market, improved a process, or learned from repeated exceptions.
The difference is critical.
An artifact proves that something happened once.
A stack proves that a capability persists.
Universities begin struggling with this distinction.
Many schools continue teaching applications: how to use spreadsheets, presentation software, learning-management systems, statistical packages, and industry tools.
A smaller group begins teaching architecture.
Students learn how to maintain personal knowledge environments, connect data sources, create verification loops, govern AI behavior, preserve project history, protect confidential information, and build reusable procedures.
The educational divide begins before graduation.
One student leaves school knowing how to operate software.
Another leaves with an operating system.
2028: The Job Breaks Into Capabilities
By 2028, businesses begin decomposing some jobs into maintained outcomes.
A small company does not necessarily need a full-time marketing coordinator. It needs three weekly posts, one newsletter, a monitored reputation system, and a monthly performance report.
It does not necessarily need a full-time researcher. It needs competitor monitoring, a Monday briefing, and immediate notification when an important condition changes.
It does not necessarily need a full-time project administrator. It needs meetings documented, commitments tracked, delays identified, and exceptions escalated.
These are pieces of jobs.
More precisely, they are capabilities.
Stack People begin assembling several of these partial demands into complete economic lives.
A 24-year-old may have one conventional job, two monthly clients, a recurring information product, and occasional project work.
The same personal stack supports all four relationships while maintaining strict separation among proprietary data.
This is when the word employment becomes less precise.
The person is not simply an employee.
The person is a small platform connecting several economic relationships.
The first employer may still provide the largest paycheck, but it no longer contains the person’s complete professional identity.
Young Stack People become more resilient than traditional employees because losing one relationship does not eliminate their entire income or professional environment.
The App Person still seeks a better job.
The Stack Person adds, removes, and improves economic relationships.
2029: The Institutions Fight Back
By 2029, established institutions recognize that employees are bringing increasingly powerful personal systems into the workplace.
Companies become concerned about confidentiality, intellectual property, cybersecurity, competing obligations, data leakage, regulatory compliance, and the possibility that important knowledge will leave with the worker.
Some organizations respond by prohibiting personal stacks.
They require employees to use company-approved applications, company-controlled models, company-owned memory, and company-defined interfaces. They treat the employee’s personal operating environment as an unacceptable risk.
These companies will still employ many people.
They will be especially attractive to App People who prefer stability, clear boundaries, institutional training, and a single governed environment.
Other companies take a different approach.
They distinguish between the company’s data and the worker’s capability.
The company owns its customer records, confidential information, proprietary processes, work product, and institutional memory.
The person retains general methods, personal learning, reusable procedures, nonproprietary agents, professional history, and the architecture of the personal stack.
The Stack Person does not refuse to conform.
The Stack Person gladly conforms to the required form.
But instead of living manually inside every company application, the employee’s stack connects through approved APIs, command-line interfaces, AI protocols, or other governed access points.
The company retains control of its system.
The worker retains architectural independence.
This is also the year the apprenticeship crisis becomes difficult to ignore.
Companies that reduced junior hiring from 2025 through 2028 begin discovering a shortage of workers with three to five years of experience.
AI preserved immediate output.
It did not automatically produce the next generation of professionals, managers, relationship holders, or institutional leaders.
The first rung must be rebuilt.
But it will not be rebuilt in its old form.
2030: The Three-Party Job
By 2030, the employment agreement increasingly has three parties:
- the institution;
- the employee;
- the employee’s operating environment.
The third party is not legally a person. It does not receive a salary or possess independent responsibility.
But it becomes functionally essential to the relationship.
The institution brings customers, capital, proprietary information, organizational authority, distribution, and a defined economic purpose.
The employee brings judgment, accountability, relationships, creativity, ethical responsibility, and lived contact with Actual.
The personal stack brings memory, prediction, procedures, agents, interfaces, accumulated methods, and the ability to maintain several streams of work.
Hiring changes accordingly.
The employer no longer asks only:
What do you know?
It asks:
What can you and your stack currently maintain?
Onboarding also changes.
Traditional onboarding teaches the person how to use the company’s interfaces.
Stack onboarding teaches the person’s system how to connect safely to the company’s environment.
The employee does not spend three weeks manually learning where information is located in ten separate applications. Approved connectors expose the relevant structure, permissions, definitions, and procedures to the worker’s synthetic subconscious.
The stack conforms to the company’s form without requiring the person to surrender the stack’s architecture.
Employment law and company policy begin addressing new questions:
Who owns an agent developed while performing company work?
Which memories must be deleted when the employee leaves?
Can the employee retain generalized procedures learned through experience?
How is client data separated across several economic relationships?
When does outside work become a conflict?
What portion of the employee’s productive system can an employer reasonably demand?
What happens when a company hires a person partly because of a stack the company does not own?
These questions reveal that the employee is no longer merely bringing labor.
The employee is bringing capital.
2031: The New Digital Class Divide
By 2031, both App People and Stack People remain common.
App companies have not disappeared. They have incorporated powerful AI, improved their interfaces, strengthened their ecosystems, and preserved large, profitable user bases.
The divide is no longer access to AI.
Both groups have access.
The divide is ownership of accumulated capability.
The App Person starts with the assigned interface.
The Stack Person starts with retained context.
The App Person uses the company’s memory.
The Stack Person connects personal memory to the company’s authorized information.
The App Person learns a workflow.
The Stack Person brings a system capable of learning workflows.
The App Person’s work is repeatedly deposited into institutional applications.
The Stack Person’s methods compound across a lifetime.
The App Person rents intelligence one interaction at a time.
The Stack Person owns the environment in which intelligence becomes increasingly useful.
This does not mean Stack People are disloyal employees.
They may become extraordinary employees.
They can integrate rapidly, understand complex environments, maintain several forms of work, and produce at a level that appears almost superhuman to organizations still structured around manual application use.
But the company cannot treat the Stack Person exactly as it treated the App Person.
It cannot reasonably demand that the Stack Person abandon a decade of accumulated digital capability and begin every morning with a blank screen.
Doing so would resemble hiring a master craftsperson and requiring the person to leave every tool, notebook, method, assistant, and practiced technique at home.
The Stack Person can follow the company’s rules.
The Stack Person can protect its secrets.
The Stack Person can use its required formats.
The Stack Person can submit to legitimate governance.
But the Stack Person will resist becoming architecturally dependent on one institution.
The person has become the platform.
Two Very Different Graduates
Consider two students graduating in 2031 with the same degree.
The first has used AI throughout college.
AI helped write papers, prepare presentations, summarize readings, create study guides, and improve applications. The student used powerful tools and received good grades.
But the work evaporated after each assignment.
The student begins the next project with a blank conversation, a blank document, and another collection of applications.
This is the App Graduate.
The second student also used AI.
But every project became part of an enduring system.
Research entered a personal knowledge environment.
Corrections became quality-control procedures.
Repeated assignments became reusable workflows.
Sources became a governed library.
Projects became evidence of maintained capability.
Client work became operating history.
Failures became updated constraints.
The student’s synthetic subconscious knows the student’s methods, vocabulary, standards, recurring weaknesses, areas of expertise, and demonstrated work.
This is the Stack Graduate.
The first graduate presents potential.
The second presents productive infrastructure.
One asks:
Will you give me an opportunity?
The other asks:
Where can this capability create value?
College Changes Last
Universities will be among the final institutions to fully accept the transition.
The university is organized around courses, departments, semesters, assignments, and applications. Each professor establishes a temporary environment. Students complete work, receive a grade, and move on.
The architecture repeatedly destroys continuity.
The student’s work accumulates on a transcript, but the student’s productive system often does not accumulate with it.
A Stack-native education would be different.
The student would leave college with more than completed courses.
The student would possess an enduring body of knowledge, a documented history of work, tested procedures, governed AI agents, verified accomplishments, trusted relationships, and a personal operating environment capable of performing real economic functions.
The degree would remain important.
But the degree would no longer stand alone.
The graduate would leave with:
a credential, a stack, and evidence.
Universities that make this transition will prepare students for the labor market that is forming.
Those that do not will continue producing highly educated App People for an economy increasingly rewarding Stack People.
The Prediction
The employment crisis facing young graduates is not simply a temporary mismatch between degrees and available jobs.
It is an early expression of a deeper architectural transition.
The traditional professional career assumed that the institution would supply the operating system. The young person would enter it, learn it, and gradually rise within it.
As the first rung weakens, young people can no longer rely entirely on that arrangement.
They will have to build part of the operating system themselves.
Underemployment will become one of the unexpected pathways into this new economy.
A job that does not use the whole person will provide income, experience, relationships, and contact with real problems. The synthetic subconscious will absorb predictable work. The personal stack will retain what is learned. Unused capacity will be directed toward additional responsibilities.
The young person will move from one underutilizing job to several productive relationships.
From underemployed to overemployed.
From application user to system owner.
From worker inside the platform to person as platform.
The great danger is that young people will continue waiting for the old doorway to reopen.
Some of it will.
Companies will eventually rebuild apprenticeships because they cannot manufacture experienced workers without first developing inexperienced ones.
But the rebuilt first rung will belong to a different ladder.
The successful young worker of 2031 will not necessarily have one perfect job.
The worker will have a direction, a portfolio of responsibilities, a synthetic subconscious, and a personal stack that becomes more capable with every completed assignment.
The App Person will continue looking for the institution capable of using all of them.
The Stack Person will know that no single institution has to.
That is the new digital class divide.
20 thoughts on “The First Stack Generation”