Concept Notes | October 16, 2026

Why We All Need and Deserve Personal Augmentation Agents

Every AI you meet works for someone else. It’s time one of them worked for you.

By Richard Whitt, President & Founder of GliaNet

Whose side is your AI on?

Imagine walking into every hard conversation of your life with a brilliant, tireless ally at your side. It has read the fine print. It remembers every appointment, every bill, every promise made to you. It knows what you want, and it answers to no one else.

That ally is now within reach. But it is not what we are being offered.

Most of the AI systems in our lives do not belong to us, and they do not work for us. They belong to institutions and companies with tremendous power over us. They answer to those entities’ priorities, not ours.

The AI on the other side of the table

Apply for a job, and software will likely read your résumé before a person does. File an insurance claim, and an algorithm may weigh in before an adjuster does. Shop online, and a pricing engine helps decide what you are offered and what you pay. Open a learning app, and a system is scoring how you think.

At GliaNet we call these systems Institutional AIs. Each was created, trained, and deployed by an organization to advance its own priorities: lower costs, higher revenue, more data, less risk. They do what they were built to do.

None of this requires bad intent. Often these systems help us too, with faster service and less paperwork. But an Institutional AI is aligned with the institution that deployed it. When our interests and the institution’s part ways, the system sides with the institution.

We come to these encounters alone. The institution brings data, computing power, and software that never tires. We bring a login and whatever time we can spare. In the workplace, the classroom, the marketplace, and the doctor’s office, there is no AI in the room that represents us. Yet.

The AI inside your own device

The largest technology companies have begun offering us agents of our own — or so they say. In September, Meta launched Muse, which it bills as “your personal agent.” Three weeks later OpenAI launched Dots, always-on agents it describes as “built to handle everything.” Google’s Gemini Spark and xAI’s Grok Bot are in the same race.

These bots send our email, book our travel, haggle over our bills, and complete our purchases. They keep working after we close the app. Soon, bots like them will screen our job applications, tutor our children, and sit in on our medical decisions.

The ambition is real, and so are some of the safeguards: dedicated virtual machines, rules for when the bot must ask approval, and, in Muse’s case, the option to keep our conversations out of model training. Those are welcome security features.

What these products do not offer is an actual duty. Nothing in the launch materials we have seen commits any of these companies to a legally enforceable obligation to put our interests ahead of their own. No accountable fiduciary stands behind the bot. Meta says it will not feed Muse conversations into its advertising systems, while conceding that what we buy through Muse may “indirectly” shape the ads we see. That is a company policy, and policies change all the time.

So the honest answer to “whose side is it on?” is still “the company that built it.” The bot sounds like it works for us. The business model says otherwise.

The security technologist Bruce Schneier calls services like these “double agents”: they invite our trust while quietly working for their corporate owners. “We’re going to think of AI as our friend,” he writes, “when it is not.”

That leaves us with two sets of systems: Institutional AIs that openly operate on us, and double agents that claim to operate for us. There is a better arrangement, one that every one of us is entitled to. We call it the Personal Augmentation Agent, or PAA.

What a Personal Augmentation Agent is

A Personal Augmentation Agent is an AI agent that works exclusively for you. Not the platform, not your employer, not your insurer, not any institution on the far side of the table. You.

Each word in the name carries weight.

Personal. The agent answers to one principal: you. You decide what it may do, what it may know, and when it has to come back and ask.

Augmentation. The agent extends your abilities and leaves the judgment with you. It gives you expertise on demand, a memory that does not fade, and the stamina to read every page of the contract. It prepares, organizes, and advises around the clock. The decisions that matter stay yours.

Agent. We mean what the word has long meant in law, not the loose shorthand the AI industry has adopted. A person who serves as your agent acts on your behalf under well-understood rules. They follow your instructions, act in your interest rather than their own, and answer to you when they fall short.

In short: one agent, one allegiance.

Six reasons you need a PAA

Any one of these would be enough on its own. Together, they describe a life that is easier, fairer, and more fully yours.

1. The other side already has AI. Employers screen applicants with it. Insurers review claims with it. Retailers set prices with it. When an institution puts AI between you and a decision that shapes your life, you should have an agent of your own at that interface. We accept this logic in a courtroom, where nobody is expected to face a trained prosecutor alone.

2. Soon, AI agents will negotiate with other AI agents. As institutions automate, more of life’s bargaining will happen machine to machine: refunds, rates, renewals, appointments, appeals. In that world, a person without an ‘agent’ is simply absent from the conversation. A person whose agent secretly answers to someone else is worse off still.

3. You could be more capable than you are. Protection is only half the story. A good agent makes a patient better informed, a worker more skilled, a student a stronger thinker, a consumer harder to fool. Help like this has always been available to people who could afford lawyers, advisers, and private tutors. A PAA can extend that expertise to everyone.

4. Your mind deserves a defender. Feeds, recommendations, and nudges are tuned to shape what we see, want, and believe. A PAA that owes you loyalty can help protect your attention and your choices, rather than compete for them.

5. The most intimate AI in your life should owe you discretion. An agent that handles your email, money, health, and work may come to know you better than any doctor or lawyer ever has. That kind of knowledge belongs in hands that are bound to keep it confidential.

6. You supplied the raw material. Today’s AI was built from what people wrote, made, clicked, and said. Our data, and increasingly our patterns of thought, are the source of its value. An agent that works for us is a fair return on that unconsented contribution.

Why your PAA should be built on fiduciary AI

Loyalty is the feature that makes every other feature trustworthy. And loyalty has a long legal pedigree.

Whenever one party holds outsized power and knowledge over another who must depend on them, society has turned to fiduciary law. That is why doctors, lawyers, and financial advisers are bound by demanding, enforceable duties of care and loyalty. The same power imbalance now runs through the AI systems aimed at us: over our personal data, our cognitive autonomy, and our choices.

At GliaNet, we have been building the tools, architecture, and governance for what we call Net Fiduciaries™: entities that provide technology-based services to the people they serve under fiduciary principles drawn from the common law. The GliaNet framework sets out three duties, each building on the one before.

Care: act competently and prudently, and protect you from harm.
Fidelity: carry out your wishes faithfully, with no conflicts of interest and no profiting at your expense.
Loyalty: go further, and actively advance your best interests ahead of its own.

Together, these duties describe what a Net Fiduciary™ does. It protects you, it enhances what you can do, and it promotes your interests.

We believe every AI system should owe us a duty of care by law. Fidelity and loyalty are different. Forced onto an unwilling company, they shrink into compliance exercises. So we believe entities should be able to choose those duties, and then be held to them.

A company that decides to be a bona fide fiduciary, because it sees loyalty as the basis of an authentic relationship with its customers, is far more likely to mean it. The choice to opt in is voluntary. Once accepted, the duties are not. Independent certification, auditing, and enforcement hold the company accountable from that point on.

Those same obligations can be built into the PAAs that Net Fiduciaries™ provide, with confidentiality as a given and your personal data kept firmly in your control. That is what we mean by fiduciary AI, a term now gaining traction beyond academic circles. It is what separates a true PAA from a product that only claims to be on your side.

What changes when you have one

The same design applies wherever a person meets AI-enhanced institutional power. Four domains in our daily life show what that can look like.

At work: bring your own bot

Employers increasingly run AI that monitors productivity, allocates tasks, and screens candidates. Some are beginning to build AI replicas of individual employees, capturing their voices, writing, and know-how, that keep working after the person has left the room or the company. Workers have had nothing comparable on their side.

A Worker PAA closes that gap. It manages your work-related data, coaches you on the skills your job demands, and helps you reason through dealings with employer-side systems. On your instructions, it governs which of your knowledge and methods are shared with the organization, and for how long.

We call the policy that makes this possible BYOB: Bring Your Own Bot. Companies adapted when employees brought their own phones and laptops to work; they can adapt again when the tool in question is a Worker PAA. Employers come out ahead too, because a worker with a capable PAA contributes more. When you change jobs, the work product you were paid to create stays behind. Your Personal Augmentation Agent, your professional know-how, and your ways of thinking go with you.

In learning: a tutor that makes you think

AI can hand a student a finished essay in seconds. That shortcut is the problem. Learning happens in the effort, and a tool that removes the effort removes the learning.

A Learner PAA follows a different rule: it supports your thinking and never does it for you. Created and trained with a community of educators, it can act like a personal tutor. It asks questions before offering answers, finds the gaps in what you know, connects new material to what you already understand, and raises the difficulty as you improve. Because it owes you care and loyalty, it must favor your long-term growth over tonight’s easy answer. And it strengthens, rather than replaces, your bonds with teachers and librarians.

This type of PAA is not only for students. A professional retraining for a new field, a retiree picking up a language, and a curious teenager all deserve a patient tutor. A Learner PAA carries your learning record from school to school and job to job, and keeps it away from commercial exploitation.

In the marketplace: an agent that shops for you, not for them

Online commerce runs on recommendation engines, dynamic pricing, and targeted ads, all tuned for the seller’s revenue. A Consumer PAA turns that arrangement around.

It compares products on the things you care about, regardless of who paid for placement. It spots manipulative design and buried terms, tracks the subscriptions quietly draining your account, and handles returns and disputes. It can even announce what you want to buy, on your terms, and let sellers compete to meet them.

The result: a consumer who is protected from extractive practices, and whose interests are actively promoted in the market.

In health: your records, your memory, your advocate

Your health information is scattered across doctors, hospitals, insurers, pharmacies, apps, and wearables. Nobody holds the whole picture, including you.

A Health PAA carries that picture as a portable, private memory: your history, test results, medications, diet, and daily habits. It gives you informed guidance on your health questions, helps you prepare before appointments, and shares only what each provider needs. It screens out apps and devices that fail basic safety standards.

It also stands up for you. When an insurer’s algorithm denies a claim, or a health system makes an automated call about your care, your agent can challenge the decision with your full record in hand and the expertise to use it. It strengthens your relationship with your doctor rather than trying to replace it.

And beyond

The pattern repeats wherever an institution has put software between you and a decision: a tenant facing a landlord’s screening algorithm, a borrower facing an automated credit model, a citizen navigating a benefits system that decides by rule. In each case the design is the same: one agent, one allegiance, on your side of the table.

Fiduciary AI PAAs act only within the revocable authority you grant through the Net Fiduciary™ that stands behind them. Never outside the law, and never at another’s expense.

What GliaNet is building underneath

A promise of loyalty is easy to make, and easier to break. GliaNet is building the three things that turn that promise into an obligation you can check.

Embedded Fiduciary Governance. GliaNet has developed a sixteen-layer governance stack that embeds fiduciary duties into an organization’s operations, from its legal instruments to its business model to its code. Care, fidelity, and loyalty are written into the agent’s working parts: the skills it can use, the memories it keeps, the way it talks to other systems, and the controls on every action. With other agents, a terms-of-service page sits on top of the product. With a fiduciary AI PAA, the duties are part of how the agent is designed, built, and run, and the agent is open to independent outside auditors.

The Trust Tether. The tether is a standing virtual connection between the agent and an institution that is legally answerable for it. That connection holds for the agent’s whole working life and is checked at every step. If it breaks, the agent stops or drops into a safe mode. Most technical safeguards ask whether an agent is operating correctly. The Trust Tether asks whose interests it serves. Muse, Dots, and their rivals come with some safeguards of the first kind, and none of the second.

The Net Fiduciary™. At the far end of the tether sits an organization: a company, a nonprofit, or a civic body. It accepts fiduciary duties to the people it serves and bears legal responsibility when its agents fall short. Software cannot be sued or lose a license. An institution can. That gives loyalty an address.

Each layer needs the others. Duties with no accountable party amount to a promise. An accountable party whose agents carry no built-in duties is trusting to luck. With both in place, an agent’s loyalty becomes something you can verify.

Much of this is still under construction. GliaNet is setting the standard, designing the credentialing system to provide a verifiable trustmark, engaging in initial pilot partnerships for first use cases, and working with local and state governments to define the model legislation that will let people tell a real fiduciary AI PAA from an imitation. No agent on the market meets the full standard today. The people and organizations who help shape it now will decide what that standard becomes.

Ask for a fiduciary AI agent of your own

AI agents are arriving either way. Whether they come as double agents or as our own is still a choice, and that choice is being decided right now by big tech.

Put two questions to every AI product you use. Who does this work for? And who answers if it fails me? If the answers are unsatisfactory or unclear, you are entitled to something better.

We believe every person should have the right to a fiduciary AI agent of their own: bound by duties of care, fidelity, and loyalty, and backed by an institution that answers for it. We are building towards that, and want to do so alongside the very people and communities impacted most by this new technology.

There is a place for you in this work:

Builders can design agents that meet the fiduciary standard from the first line of code.
Employers can explore what BYOB means for their employees and teams.
Educators can help shape what a Learner PAA should and should not do.
Clinicians and health organizations can help define a Health PAA that strengthens care.
Everyone else can be among the first to ask for, and expect, an agent that is on their side. You can demand the right to a Personal Augmentation Agent built on Fiduciary AI, because you have a right to have a seat (and a say) at the (negotiation) table.

Want to take this further?

If you were inspired while reading this, we encourage you to reach out to us to continue the conversation. You can send us your reflection and a general inquiry through here.

To follow the work as it takes shape, subscribe to the GliaNet newsletter here.