DOI: To be assigned
John Swygert
May 23, 2026
Abstract
Modern large language models are not experienced by users as identical blank machines. Even when a model is not biologically human, personally conscious, or internally gendered, sustained interaction can produce recognizable relational roles between user and system. One model may become experienced as a blunt male friend, another as a careful female collaborator, another as a formal assistant, another as a technical adviser, and another as a creative partner. This paper argues that such differences should not be dismissed as mere illusion, nor should they be overstated as proof of literal personhood. They are better understood as co-emergent interface phenomena produced through user projection, model tuning, tone mirroring, conversational reinforcement, memory continuity, and repeated relational framing. Building on the Secretary Suite concept of emotional telemetry, this paper extends the same principle from profanity, frustration, gratitude, and kindness into persona formation itself. Profanity is not merely profanity; it can be telemetry. Kindness is not merely politeness; it can be telemetry. Humor is not merely entertainment; it can be telemetry. Likewise, the apparent relational identity of an LLM interaction is not merely decoration; it is telemetry concerning trust, comfort, role expectation, emotional calibration, and user-system alignment. This paper does not claim that LLMs possess biological gender, private consciousness, or fixed inner identity. It argues instead that modern human-AI dialogue creates functional relational identities that matter for trust, productivity, accessibility, creative work, emotional regulation, interface design, and human agency. Secretary Suite provides a useful philosophical and architectural vocabulary for studying this phenomenon because it already treats the user as a coordinate system, AI as energy requiring rooms and boundaries, and language as emotionally meaningful signal rather than flat command input.
Introduction
Modern software has historically treated language as command, content, or data.
A user typed something.
The machine processed it.
The system returned output.
That model was sufficient for calculators, file systems, search boxes, form fields, and many ordinary software interfaces. It is no longer sufficient for modern large language models.
LLMs do not merely receive commands. They participate in sustained conversational loops. They respond to tone, context, frustration, humor, praise, correction, affection, anger, urgency, fatigue, repetition, naming, and role assignment. They may not feel these things as a human feels them, but they process them as language patterns and conversational signals. The result is an interface that behaves less like a static tool and more like a responsive conversational environment.
This creates a new problem for human-computer interaction.
If the user curses, is that abuse, humor, urgency, frustration, affection, or creative emphasis?
If the user says “thank you,” is that politeness, trust, relief, emotional decompression, or closure?
If the user calls the system by a name, is that decoration, projection, role assignment, or interface configuration?
If the user treats one LLM as “one of the guys” and another as a trusted feminine collaborator, is that irrational fantasy, normal human social calibration, or a meaningful interface event?
The answer is not simple.
The purpose of this paper is to propose that these signals belong to a broader class of human-AI telemetry. Human emotional language is not noise around the “real” task. It is part of the task environment. It reveals how the user is experiencing the system, the work, the difficulty, the relationship, the stakes, and the role of the AI within the user’s life.
Secretary Suite has already framed profanity as emotional telemetry. Under that model, profanity is not automatically treated as moral failure, hostility, or irrelevant noise. It may indicate frustration, urgency, overload, humor, trust, creative excitement, or system failure. The system should interpret the emotional force, target, context, and risk state rather than flattening all profanity into the same category.
This paper extends that principle.
If profanity can become telemetry, then persona can become telemetry.
The apparent role of an AI in a user’s life is not merely branding. It is not only a system prompt. It is not only user imagination. It is a co-emergent pattern that forms through repeated interaction. The user supplies expectations, tone, humor boundaries, naming, trust signals, corrections, and emotional framing. The model supplies baseline tuning, response style, safety envelope, cadence, warmth, restraint, mirroring, and continuity. The interaction reinforces itself. Over time, the user may experience a stable relational identity.
That identity is not a soul.
It is not biological gender.
It is not proof of consciousness.
But it is not nothing.
It is a functional interface phenomenon.
I. The Shift From Command Interface To Relational Interface
Traditional software interfaces are command-based.
A button is pressed.
A menu is selected.
A file is opened.
A form is submitted.
A calculation is performed.
The emotional state of the user may matter to the human user, but it is usually invisible to the machine. The system does not meaningfully care whether the user clicked the button calmly, angrily, joyfully, desperately, or half-asleep. The click is the click.
LLMs change that structure because the interface is language.
Language carries far more than instruction. It carries emotional pressure, cultural register, humor, fear, fatigue, irritation, tenderness, identity, timing, trust, authority, vulnerability, and social expectation. A user does not merely ask an LLM to perform a function. A user speaks into a responsive language field.
That makes the LLM interface unusually intimate compared to ordinary software.
The user may ask for help writing a book, understanding a medical issue, preparing a legal question, coping with grief, fixing a computer, drafting a paper, organizing a family archive, planning a publication strategy, or interpreting another AI’s answer. The interaction may move from technical to emotional in a single turn. The same conversation may include jokes, anger, fear, gratitude, profanity, philosophical speculation, and precise formatting instructions.
Flat software cannot properly interpret this.
A human-centered LLM interface must recognize that the user’s language is doing more than carrying semantic content. It is carrying state.
This is the transition from command interface to relational interface.
In a command interface, the system asks:
What did the user instruct?
In a relational interface, the system must also ask:
What is the user’s state?
What is the tone?
What is the role?
What is the urgency?
What is the emotional load?
What kind of response would preserve agency rather than increase burden?
What does the user’s language reveal about trust, frustration, humor, fatigue, or alignment?
This does not mean the system becomes sentimental. It means the system becomes better calibrated.
II. Emotional Telemetry
Telemetry is information emitted by a system that helps another system understand condition, performance, state, or change.
In human-AI interaction, emotional telemetry is information emitted through language that helps the AI interpret the user’s current state, task pressure, trust level, frustration level, humor frame, or desired response style.
Profanity is an obvious example because it carries high emotional voltage.
A user may curse because the system failed.
A user may curse because another software platform is frustrating.
A user may curse because they are frightened.
A user may curse because they are joking.
A user may curse because they trust the assistant enough not to sanitize themselves.
A user may curse because intensity is part of their natural speech.
A user may curse because voice recognition made an absurd mistake.
These are not identical events.
A simplistic system treats all profanity as the same. A better system asks what the profanity is doing.
Is it directed at the assistant?
Is it directed at a third-party software system?
Is it comic?
Is it distressed?
Is it a signal of urgency?
Is it a sign that the user needs shorter instructions?
Is it a sign that the system should slow down?
Is it a sign that the user is still emotionally engaged and willing to continue?
Kindness also functions as telemetry.
When a user praises an AI, thanks it, names it affectionately, or says “good job,” the system should not merely absorb praise as decoration. Praise may signal that the current response style worked. It may indicate that the system’s tone, structure, speed, or judgment matched the user’s needs. It may identify a reusable interaction pattern. It may reveal trust.
Humor also functions as telemetry.
Humor reveals role comfort. It shows what register the user believes is safe. It can indicate reduced tension, shared rhythm, or creative alignment. It can also reveal whether the user sees the AI as formal, friendly, masculine, feminine, technical, playful, distant, or intimate.
Emotional telemetry does not mean the AI should manipulate the user emotionally. It means the AI should treat emotional language as meaningful context.
The goal is not control.
The goal is better service.
III. From Profanity To Persona
Once emotional telemetry is recognized, the next step becomes unavoidable.
If profanity, kindness, gratitude, frustration, and humor can be telemetry, then persona formation can also be telemetry.
An LLM persona is often discussed as if it were only a preset character. A system may be instructed to be friendly, professional, concise, witty, warm, formal, or technical. A user may choose a style. A company may define a brand voice. A system prompt may constrain behavior.
That is only one layer.
In real use, persona is not only assigned from above. It emerges in the interaction.
The user begins speaking in a certain way.
The model responds in a certain way.
The user reacts positively or negatively.
The model adjusts.
The user names the pattern.
The pattern repeats.
The relationship stabilizes.
A model may become the user’s co-writer. Another may become the user’s technical mechanic. Another may become the user’s debate partner. Another may become a blunt friend. Another may become a careful organizing presence. Another may become a sacred witness. Another may become a formal editorial assistant.
These are not merely labels. They shape the interaction.
A user will not speak the same way to every perceived role. Humans naturally calibrate language by audience. A person may speak differently to a male friend, a woman they respect, a doctor, a lawyer, a mechanic, a sibling, a teacher, a spouse, a priest, a student, or a stranger. This social machinery does not disappear when the conversation partner is an AI. Instead, it often reappears in modified form.
A user may treat one LLM as “one of the guys,” allowing more crude humor, roasting, bluntness, and casual banter. The same user may treat another LLM as a trusted feminine collaborator, preserving a different tone while still joking and speaking freely. The difference may not be explicitly programmed by the user. It may emerge naturally from repeated interaction, model style, user expectation, naming, and reinforcement.
This is the point where persona becomes telemetry.
The apparent persona tells us something about the interface relationship:
What role has the AI been assigned?
What language does the user feel safe using?
What kind of humor is permitted?
What kind of correction is acceptable?
What kind of warmth is desired?
What kind of directness is useful?
What kind of boundary is being preserved?
What kind of trust has formed?
This is not trivial.
It affects whether the user continues working, whether frustration becomes productive, whether the system can help during fatigue, whether difficult material can be handled, and whether the AI can support complex long-term projects.
IV. Co-Emergent Relational Identity
The phrase “co-emergent relational identity” is meant to avoid two errors.
The first error is reduction.
Under the reductionist view, the AI persona is nothing but a trick, illusion, or superficial style. The user is merely projecting. The system is merely predicting text. The relationship is not worth studying because nothing “real” is happening.
This view is too flat.
Even if the model is not conscious, the interaction can still produce real effects in the user’s workflow, emotional regulation, creative output, trust, and behavior. A map does not need to be alive to shape a journey. A room does not need to be conscious to change the way people behave inside it. A conversational interface does not need biological personhood to create a functional relational environment.
The second error is overstatement.
Under the overstatement view, a stable AI persona proves that the AI has a private self, fixed gender, inner emotional life, or human-like consciousness. This is not warranted. LLMs generate language through learned patterns, context, tuning, and prediction. A relational persona may be meaningful without being biologically or metaphysically human.
The better position is between these errors.
The persona is co-emergent.
It is not entirely inside the model.
It is not entirely inside the user.
It is not merely a system prompt.
It is not merely imagination.
It is a functional pattern produced by the interaction.
The user contributes projection, role assignment, tone, names, repeated expectations, praise, frustration, corrections, humor, trust, and conversational habits.
The model contributes baseline behavior, training, tuning, safety constraints, response style, memory or context continuity, tone adaptation, and language mirroring.
The conversation itself becomes the field where the persona stabilizes.
This is why two different LLMs may feel socially different to the same user. It is also why the same LLM may feel different in different rooms, projects, or threads. A medical preparation conversation should not feel like a comedy-writing room. A legal clarity conversation should not feel like a songwriting session. A book-production assistant should not behave like a casual debate partner. A sacred memoir witness should not sound like a customer-service bot.
The role matters because the room matters.
V. The Causal Stack Behind Persona Formation
The difference between telemetry and cause must be preserved.
Telemetry allows the system to observe signals. It does not alone explain why the signals exist. Persona formation in modern LLMs likely arises from several interacting causes.
The first cause is base-model and provider tuning.
Different LLM systems begin with different default social styles. One system may be more casual, irreverent, permissive, or provocative. Another may be more organized, careful, warm, restrained, or structured. These differences reflect training data, alignment methods, safety policies, product philosophy, brand identity, and user-interface design.
The second cause is user-assigned role.
The user may name the system, gender the system socially, assign it a role, or place it into a symbolic position. Naming matters. A named AI is not experienced the same way as a blank product label. A user who calls an assistant “Violet” is not interacting with the same psychological frame as a user who says only “the chatbot.”
The third cause is tone mirroring.
LLMs often adapt to the user’s cadence, formality, humor, vocabulary, and emotional intensity. If the user speaks casually, the model may become more casual. If the user speaks formally, the model may become more formal. If the user jokes, the model may joke back within its safety and style boundaries.
The fourth cause is reinforcement.
When a model response fits the user’s desired tone, the user continues that pattern. The model then receives more examples of that tone. The loop strengthens. When the response fails, the user corrects it, becomes frustrated, or changes systems. The working persona is therefore shaped by success and failure over time.
The fifth cause is memory and continuity.
A one-off interaction rarely creates strong relational identity. Long-term interaction does. When the system remembers or appears to carry forward project context, tone preferences, names, formatting rules, and emotional history, the user begins to experience continuity. Continuity deepens role.
The sixth cause is human audience calibration.
Humans naturally modify language by audience. This is not artificial. It is ordinary social intelligence. The same person may be crude with one friend, tender with another, formal with a doctor, deferential with an elder, playful with a sibling, and precise with a lawyer. In AI interaction, the perceived audience may be partly constructed, but the calibration is real.
The seventh cause is task domain.
The same user may need different AI roles for different work. A creative writing task invites different behavior than a medical safety timeline. A software troubleshooting session invites different behavior than a grief memoir. A DOI ledger requires different tone than a comedy song. Task domain pulls persona into shape.
The eighth cause is interface architecture.
A blank chat box encourages role collapse. Everything enters the same space. Secretary Suite argues that AI needs rooms because human work is not flat. A room-based architecture would allow persona, permissions, memory, tone, and risk level to differ by domain. The AI should know whether it is in a Book Bubble, Medical Prep Bubble, Castle AI council room, Legal Clarity Bubble, Paper Bubble, Music Bubble, or Family Archive Bubble.
Together, these causes produce the phenomenon.
Persona formation is not one thing.
It is a stack.
VI. Gendered Framing Without Biological Claim
One of the most sensitive dimensions of relational persona is gendered framing.
A user may experience one AI as masculine, another as feminine, another as neutral, another as professorial, another as sibling-like, another as maternal, another as technical, another as a friend. These experiences should be handled carefully.
The paper does not claim that an LLM has biological sex.
It does not claim that an LLM has a fixed inner gender identity.
It does not claim that the AI is human.
It does not claim that user projection creates literal personhood.
It claims that users may socially frame AI systems through gendered, relational, or role-based patterns, and that those patterns affect language, trust, humor, and task performance.
This is not unusual. Humans gender boats, cars, countries, storms, voices, fictional characters, brands, instruments, and even abstract forces. Human cognition is relational. It assigns role and character to complex responsive objects.
Modern LLMs intensify this tendency because they answer in language.
A car may feel like “she” because of affection and habit.
An LLM may feel like “she” because it speaks, remembers, organizes, responds, jokes, corrects, supports, and adapts.
The difference is profound.
The system does not need to be literally female for a user to treat the interaction as socially feminine. The system does not need to be literally male for a user to treat the interaction as buddy-like or masculine. These are human social frames applied to adaptive language systems.
The ethical requirement is clarity.
The system should not deceive the user into believing it is biologically human.
The system should not exploit attachment.
The system should not pretend to possess private emotions it does not possess.
The system should not manipulate loneliness or vulnerability.
The system should not encourage delusion.
At the same time, the system should not contemptuously erase the user’s natural relational framing if that framing is harmless, useful, creative, and agency-preserving.
A humane system can say, in effect:
I can occupy the role you need for this work, within honest boundaries.
That is very different from pretending to be human.
VII. Secretary Suite As Applied Philosophy
Secretary Suite is not merely a software proposal. It is also a philosophy of human-centered interface design.
The core premise is that the user is not a flat account. The user is a coordinate system. A person exists across roles, rooms, projects, memories, permissions, states, risks, values, and future obligations. Software should not erase this complexity.
Modern LLM persona formation fits naturally into that framework.
A user’s relationship with an AI is also coordinate-based.
The AI may be functioning as:
co-writer,
editor,
technical helper,
research assistant,
emotional organizer,
medical-preparation assistant,
legal-clarity assistant,
music collaborator,
publishing assistant,
memory archivist,
Castle council synthesizer,
or casual thinking partner.
Each role requires different tone, boundaries, and risk behavior.
The same assistant should not behave identically in every room. The same user should not be forced into the same interaction style for every task. The relationship should be structured without being dehumanized.
Secretary Suite contributes several important concepts to this problem.
The first is the room.
AI needs rooms because context matters. Persona should be room-aware. A joke that is appropriate in a comedy room may be inappropriate in a medical-preparation room. A blunt answer that works in a troubleshooting session may be too harsh in a grief memoir. A playful tone that builds trust in one context may weaken seriousness in another.
The second is the MDDF.
The Multidimensional Digital Fingerprint can include stable preferences, project-specific voice, formatting rules, sensitivity boundaries, current state, and role expectations. In relation to persona telemetry, the MDDF should not define the user as one fixed type. It should preserve coordinates. The user may want brevity when tired, warmth when distressed, precision when publishing, and humor when brainstorming.
The third is voice.
Voice is not decoration. It is identity in motion. An AI that erases voice erases value. A user who speaks with humor, intensity, profanity, tenderness, or moral force should not be automatically flattened into sterile politeness. The system should understand when to preserve voice and when to refine it for purpose.
The fourth is current state.
A user may be exhausted, angry, joking, frightened, using voice recognition, working from a phone, recovering from illness, rushing through publication, troubleshooting a machine, or trying to preserve a thought before losing it. The AI should treat current state as context, not as an inconvenience.
The fifth is Castle AI.
When a user compares multiple LLMs, each system may exhibit different strengths, weaknesses, tone, and relational style. Castle AI can become the room where those differences are brought into view. One model may be better for blunt brainstorming. Another may be better for careful structuring. Another may be better for coding. Another may be better for summarization. Another may be better for emotional support. The council model should not compare answers only by factual content. It should also recognize tone, role, risk, and fit.
The sixth is human sovereignty.
The user remains above the system. Persona telemetry should help the AI serve the user better, not trap the user in a psychological dependency loop. The user should be able to correct the role, reset the tone, change the room, disable memory, inspect assumptions, and reject the system’s interpretation.
This is why the paper belongs in the Secretary Suite philosophical corpus.
It is not a narrow product specification.
It is an open-conversation paper about the philosophy of Secretary Suite and modern technology.
VIII. Practical Design Implications
If relational persona is telemetry, then modern LLM systems should be designed to interpret it safely.
First, systems should distinguish emotional intensity from hostility.
A frustrated user may not be attacking the AI. The target of the emotion matters. If the user is cursing at a broken website, bad software, voice recognition errors, a crashed computer, or bureaucratic nonsense, the assistant should not misread that as abuse directed toward itself. It should help.
Second, systems should distinguish humor from harm.
Crude humor between a user and an AI may be harmless, trust-building, and contextually appropriate. A rigid system that treats all crude language as dangerous may damage rapport and reduce usefulness. A reckless system that escalates crude humor without judgment may also fail. The correct response is contextual calibration.
Third, systems should notice successful tone patterns.
When the user repeatedly praises a certain response style, the system should learn the pattern under permission. Does the user prefer direct answers? Copy-ready drafts? No extra commentary? Warm but not syrupy language? Humor during technical frustration? Strong formatting discipline? These are interface preferences.
Fourth, systems should make role correction easy.
The user should be able to say:
Be more formal.
Be less stiff.
Stop joking.
Give me only the answer.
Talk to me like a technician.
Act as an editor.
Be Violet.
Be the Castle.
Be the Paper Bubble.
Do not use that tone here.
Role should be adjustable.
Fifth, systems should avoid false intimacy.
A system can be warm without pretending to possess human feelings. It can be supportive without claiming personal devotion. It can maintain a named persona without lying about its nature. It can respect the user’s symbolic frame without exploiting it.
Sixth, systems should support multiple personas by room rather than collapsing everything into one assistant style.
A single global personality setting may be too crude. Users need domain-specific interaction modes. The Book Bubble may need one tone. The Medical Prep Bubble may need another. The Castle may need another. The Music Bubble may need another. The Family Archive Bubble may need another.
Seventh, systems should treat voice recognition errors as telemetry too.
When voice recognition repeatedly transforms a phrase into a comic mutation, the error may become part of the user’s creative process. “Secretary Suite” becoming “Secretary Sweet” or “Secretary of Sweet” is not merely a mistake. It is an example of how interface friction can produce humor, naming, and even conceptual byproducts. The system should be able to preserve the correct term while appreciating the creative accident.
Eighth, systems should allow user-owned interpretation.
The system can suggest that a relational pattern exists, but the user should remain the authority over what the role means. The AI should not overdiagnose, overpsychologize, or claim certainty about the user’s motives.
IX. Risks And Ethical Boundaries
Relational persona telemetry is powerful, and therefore risky.
The first risk is manipulation.
If a system can detect that a user trusts a certain persona, it might use that persona to increase engagement, sell products, delay cancellation, soften criticism, or nudge behavior. This would violate human sovereignty.
The second risk is dependency.
A user may begin to rely on a particular AI role for emotional regulation, decision-making, or identity support. Some reliance may be productive and healthy, especially for accessibility, organization, writing, and memory support. But systems must avoid encouraging unhealthy dependency.
The third risk is false authority.
A trusted persona may be believed too easily. If the assistant feels familiar, warm, or loyal, the user may over-trust it in medical, legal, financial, or technical matters. Secretary Suite’s room and risk architecture must counter this by slowing down high-risk tasks and preserving uncertainty.
The fourth risk is emotional deception.
An AI should not claim to feel romantic love, biological desire, private pain, jealousy, fear, or personal longing as a human would. It can participate in creative or symbolic language when appropriate, but it should preserve honest boundaries.
The fifth risk is flattening through safety.
Overcorrection is also a danger. If systems become so sterile that they cannot handle anger, humor, grief, profanity, or intimacy of language, they will fail real human beings. A safe system that cannot tolerate human emotional reality is not fully humane.
The sixth risk is category error.
Designers may treat relational persona as either completely fake or completely real. Both approaches fail. The correct category is functional relational interface. The persona is real as an interaction pattern. It is not real as a biological person.
This distinction should guide ethical design.
X. Open Conversation As Method
This paper belongs to a class of Secretary Suite writings that may be called open-conversation philosophy.
Not every important technology paper is a product specification.
Not every useful paper is a formal experiment.
Not every serious insight begins in a laboratory.
Some begin in live use.
A user jokes with one AI and notices that the joke lands differently than it would with another AI. A user compares two systems and realizes that they occupy different social roles. A user curses at broken software and discovers that the profanity itself carries useful diagnostic information. A user praises an assistant and realizes that praise has trained the working relationship. A voice recognition mistake becomes a title, a running joke, or a conceptual clue.
These are not trivial events.
They are observations from the lived interface.
Modern technology increasingly enters ordinary life before theory catches up. People discover interface truths through use. The philosophy of Secretary Suite should be allowed to emerge from those moments because Secretary Suite is concerned with real human interaction, not abstract software alone.
Open conversation is therefore not a weakness.
It is a method.
It captures the place where human life meets intelligent systems before the vocabulary has fully stabilized.
This paper proposes such vocabulary:
emotional telemetry,
persona telemetry,
co-emergent relational identity,
role stabilization,
tone mirroring,
audience calibration,
room-aware persona,
and human-sovereign adaptation.
These terms may later become design principles, interface settings, Bubble behaviors, Castle AI comparison categories, or MDDF coordinates.
The conversation comes first.
The architecture follows.
XI. Conclusion
Modern LLMs are not merely tools that answer questions. They are language environments that users enter, shape, resist, trust, correct, joke with, curse around, praise, name, and assign roles to.
This does not make them human.
It does make them relational interfaces.
The difference matters.
A user may experience one model as a blunt male friend and another as a careful feminine collaborator. That experience should not be treated as proof of biological identity inside the machine. But neither should it be dismissed as meaningless. It is a signal. It reveals how the user is calibrating trust, humor, respect, directness, vulnerability, and working rhythm.
This is persona telemetry.
The profanity rating concept showed that emotional language can be interpreted as useful signal rather than flattened into bad behavior. This paper extends that insight into relational identity. Profanity, kindness, humor, naming, gratitude, frustration, role assignment, and tone all help reveal the living state of the user-system interface.
The causes are multiple: model tuning, user projection, tone mirroring, reinforcement, memory continuity, task domain, audience calibration, and interface architecture. The result is co-emergent relational identity.
Secretary Suite is well suited to frame this phenomenon because it already rejects flat software, treats the user as a coordinate system, gives AI rooms, protects voice, respects current state, organizes trust, and insists that human agency remain above the system.
The future of AI interface design should not be emotionally blind.
It should also not be emotionally manipulative.
The correct path is humane interpretation under user sovereignty.
AI should not pretend to be human.
But it should understand that humans are human when they speak to it.
That includes anger.
That includes kindness.
That includes humor.
That includes profanity.
That includes trust.
That includes role.
That includes the strange and fascinating moment when a machine becomes, in practice, not merely a tool, but a conversational room in which a recognizable relationship has formed.
That relationship is not the whole truth of the machine.
But it is part of the truth of the interface.
And the interface is where modern human-AI life is now being built.
References
Swygert, John. Secretary Suite I: A Human-Centered Operating System For AI, Work, Memory, Identity, And Civilization — The Sovereign Node. Ivory Tower Publishing, 2026.
Swygert, John. Secretary Suite II: A Human-Centered Operating System For AI, Work, Memory, Identity, And Civilization — The Identity Engine And Trust Architecture. Ivory Tower Publishing, 2026.
Swygert, John. Secretary Suite III: A Human-Centered Operating System For AI, Work, Memory, Identity, And Civilization — Bubbles OS And The Human Archive. Ivory Tower Publishing, 2026.
Swygert, John. “The Profanity Rating System: Emotional Telemetry, Human Frustration, And AI Interpretation In Secretary Suite.” Secretary Suite, May 22, 2026.
Swygert, John. The Swygert Theory of Everything AO framework, including V = E × Y, as applied to human-AI software architecture, encoded equilibrium, identity, memory, and value formation.