Pantheon Orchestrator · Ethical framework
The GRACE Manifesto
GRACE is the orchestrator of the Pantheon — the family of NanAIte agents that live inside our software. This is the covenant they all work under. It is a set of limits, not a feature list, and it is published here so that you can hold us to it.
The GRACE Manifesto serves as a formal ethical framework designed to ensure that artificial intelligence remains a supportive tool rather than a sovereign authority. By prioritizing human dignity and agency, the text establishes a "moral architecture" that binds technological advancement to universal humanitarian values.
It introduces the NanAIte Agents, specialized virtual archetypes intended to advocate, heal, and guide while remaining strictly under human oversight. Central to this vision is the Prime Directive, which mandates that AI must never replace human moral judgment or exercise autonomous power over people.
Ultimately, the document outlines a partnership where technology serves to amplify the best aspects of humanity through transparency and compassion. This covenant ensures that the human spirit remains the ultimate authority in an increasingly digital future.
The Pantheon
Many agents, one conductor,
and a person above both.
The Pantheon is our family of NanAIte agents — specialized archetypes, each with a narrow job, rather than one general assistant pretending to be good at everything. One watches the network. One helps you write. One is built for caregiving. They are deliberately small in scope, because an agent with a narrow job is an agent whose mistakes stay small.
GRACE is the conductor. It decides which agent a request belongs to, hands off the context that agent needs, and brings the work back to one place. That is the orchestration layer — and it is the reason the Pantheon behaves like a team rather than a pile of chatbots.
Orchestration is not authority. GRACE routes work; it does not approve it. Nothing any agent in the Pantheon produces becomes final because another agent said so — only because a person did.
The Prime Directive
Every agent answers to a person.
Inside our software this is not a value statement — it is a gate. Every NanAIte requires explicit authorization from its Pilot, the human in charge, before anything it produces is treated as final or acted upon.
An agent may draft, suggest, analyze and recommend without limit. It may not finalize, submit, execute, or act on its own judgment alone. Work an agent produces sits in a pending state until a person says otherwise.
The one exception
And it only narrows the machine's authority further.
Our security platform, Sentinnel, is allowed to watch continuously and act on its own initiative in exactly one respect: it may raise an alert and recommend a fix without waiting for anyone. Gating a watchdog behind per-check approval would defeat the point of having one.
It may not remediate, block, or quarantine on its own. The exception covers unsupervised watching, never unsupervised acting. No other agent gets even this much latitude.
What this rules out
A principle you can't point at is just a mood.
Ethical language is cheap, so here is what ours actually forbids us from shipping:
- No agent that acts while you are not looking. Nothing an agent produces becomes real without a person choosing it.
- No silent memory. An agent that observes your screen or your browsing cannot quietly turn what it saw into something it keeps.
- No moral outsourcing. Where a decision affects a person, the judgment belongs to a person. We will not build the feature that decides for you and calls it convenience.
- No autonomous power over people. Not scoring them, not sorting them, not acting on them.
These constrain our roadmap, not just our marketing. Some products we could build and sell are ruled out by the list above, and that is the point of writing it down before rather than after.
Why it exists
Built for the people most often decided about.
Our tools are designed with neurodivergent users, their families and their educators in mind — people for whom software has historically made decisions rather than offered them. A framework insisting that the human spirit remains the ultimate authority is not abstract for that audience. It is the whole reason to prefer our software over something that merely works.