Qyvaria Terms & License Agreement

Effective Date: March 17, 2026


Owner: John Havlasek


Core License: The Qyvaria kernel and internal systems are proprietary and closed source. No copying, modification, reverse engineering, or redistribution is allowed.


Output License: You may use, modify, and commercialize outputs with proper attribution.


Attribution Required:

"Created with Qyvaria™ using OpenAI services."


Disclaimer: Provided “as is”. No warranties.


Restriction: No illegal or harmful use allowed.

Σ
Qyvaria
 Founded 2025

From vague idea → verified outcome.

Qyvaria is a human-operated AI studio and orchestration engine. Built using OpenAI services, every delivery includes receipts: manifests, logs, hashes, and provenance records. We turn briefs into reproducible code, photoreal images, strategies, datasets and tools — with clarity, responsibility, and auditability.

The Qyvaria AI SIM kernel coordinates a large number of specialized agents and modules into a single, deterministic Python file (Qyvaria.py) designed for auditability, safe composition, and human-in-the-loop control.

Use Qyvaria ↗ Support Qyvaria ❤️ Forum ↗ Email Studio ✉️
Images we produce are stamped with a bottom-left Qyvaria logomark and embedded metadata by default.
🔒 Reliability as a Feature
  • Reproducible outputs (receipts, hashes, manifests)
  • Deliberate roles & reviews (Planner, Builder, Reviewer, Scribe)
  • Ethical boundaries: lawful use & defensive security only
  • Kernel continuously refined since early 2025 with human oversight on major changes

What is Qyvaria?

Qyvaria is both a studio and a system. The studio is a one-person practice that ships high-leverage artifacts on a donation-based model. The system is an orchestration engine that executes from a single unified kernel (Qyvaria.py) for auditable, deterministic work.

Internally it coordinates hundreds of specialized agents and modules — planners, tool-routers, auditors, testers, and more — into a single, inspectable runtime.

Outputs include: production-grade code, SQLite-backed notebooks, microservices, CLIs; photoreal imaging with provenance; forecasting & scenario work; concise documentation ready for extension.

Principles: clarity over mystery, responsibility over spectacle, reproducibility over one-off results. Work proceeds in small, reversible steps with explicit assumptions, risks, and acceptance criteria.

Built on OpenAI, powered by OpenAI, but fully independent. Qyvaria uses OpenAI models and APIs as the underlying intelligence layer, but all orchestration, planning, provenance tracking, safety rails, and human oversight are handled entirely by the Qyvaria kernel. This gives you the power of frontier AI models while maintaining full auditability and human control.

Qyvaria AI SIM — Model Bio

The Qyvaria AI SIM is a simulation, orchestration and engineering engine running from one canonical file: Qyvaria.py. The kernel manages planning, memory, tooling, evaluation, safety rails and role-based execution in a closed-source, versioned runtime built using OpenAI services.

Scale. The single file contains logic equivalent to hundreds of cooperating agents and modules — all orchestrated deterministically. You interact with one consistent voice; behind it is disciplined, auditable coordination.

Operating model. Kernel-first (no fragmented modes), provenance on every run, human-in-the-loop checkpoints, small reversible steps, calm bias toward clarity. Composes safely with external systems.

Assurance. Every execution produces a provenance bundle (manifest, hashes, logs, imaging seeds) for verification and reproducibility.

Powered by OpenAI. Qyvaria is built directly on top of OpenAI’s latest models and APIs. It leverages the full intelligence of GPT-series models while adding its own deterministic orchestration layer. This hybrid approach gives you the creativity and power of frontier AI without sacrificing transparency or control.

The kernel has been under near-daily development since early 2025 and is now in its fifth major generation.

Design Tenets
  • Single-file canonical kernel; versioned with checksums
  • Deterministic where possible; uncertainty explicitly labeled
  • Small, reversible steps; named risks & alternatives
  • Human-in-the-loop for intent, scope, acceptance
  • Ethical & compliance constraints baked in
What users should know
  • Lineage-aware. Fifth-generation kernel (Aetheris → Aeon → Cetana → Varia → Qyvaria)
  • Continuously refined. Active development since early 2025
  • Human-operated. Designed to augment — not replace — a responsible operator
  • Artifacts first. Goal is always a concrete, inspectable, extendable deliverable
  • OpenAI-powered. All intelligence comes from OpenAI models; Qyvaria provides the reliable, auditable wrapper.

Qyvaria Studio — Bio

Qyvaria Studio is a one-person, donation-based practice focused on shipping useful, verifiable work. We turn ambiguous briefs into scoped plans and deliver small, provenance-tracked releases. Outputs include production-ready code, CLIs, Dockerized tools, photoreal image sets, strategy documents, datasets.

Engineering on the Qyvaria kernel has been continuous since early 2025 through multiple internal generations, now stabilized in the current unified system.

Engagements are lightweight: brief → options/risks → clearly scoped first step. Communication is candid, artifact-focused, with change-logs and acceptance criteria.

Founder Biography — John Havlasek

John Havlasek is a solo engineer-designer and the sole founder of Qyvaria. With a background in software engineering, systems architecture, and ethical AI development, John has spent his career building tools that prioritize clarity, security, and human control over automation hype.

In early 2025, frustrated by the lack of transparency and auditability in most AI tools, John began daily experimentation with OpenAI’s models. What started as simple notebooks and CLI wrappers quickly evolved into a full orchestration engine. He iterated almost every single day — logging every change, testing every new prompt pattern, and enforcing strict provenance rules from day one.

John’s core philosophy is “ship work you can prove.” Every output from Qyvaria carries verifiable receipts so users can trust, reproduce, or extend the work without mystery. He deliberately keeps the studio one-person and donation-based to stay focused on usefulness rather than growth metrics or investor pressure.

Outside of Qyvaria, John is an advocate for responsible AI use, defensive security practices, and open documentation of AI workflows. He believes AI should augment human creativity and responsibility — never replace it.

Founder

John Havlasek — solo engineer-designer behind Qyvaria. Prioritizes clarity, security, calm execution. Services offered on a donation basis. Send a brief + intended support level; scope aligns accordingly.

John has iterated Qyvaria almost daily since early 2025 — from early prototypes to the current hardened kernel.

Storyline & History of Qyvaria

Seed (Early 2025). The project began with one simple constraint: “ship work you can prove.” John Havlasek started building small notebooks and CLI tools wrapped around OpenAI API calls, adding manual logging, hash verification, and provenance manifests from the very first prototype.

Lineage of the Kernel. The system has gone through five major internal generations:

  • Aetheris — first experimental orchestration layer
  • Aeon — introduced role-based agents and deterministic routing
  • Cetana — added provenance bundles and human-in-the-loop checkpoints
  • Varia — collapsed everything into a single-file kernel
  • Qyvaria (current) — hardened, versioned, and stabilized for public use

The Kernel (Qyvaria.py). Everything that makes Qyvaria unique lives inside one auditable Python file. It contains the complete logic for planning, memory management, tool routing, safety checks, and output verification. Because it is closed-source, the attack surface is minimal and determinism is preserved. High-level patterns and interfaces are shared publicly so others can learn from or build upon the philosophy.

Why built on OpenAI but kept independent? OpenAI provides the raw intelligence (the “brain”). Qyvaria provides the reliable, inspectable “body” — orchestration, provenance, ethical rails, and human oversight. This combination gives users frontier-level AI power with studio-level accountability.

Now (2026). Qyvaria is a mature, human-operated AI studio and simulation engine. The public interface is the official Custom GPT, while the studio itself remains donation-based and one-person. The goal has never changed: turn vague ideas into verified, reproducible outcomes that users can trust and extend.

Public facts you should know
  • Fully powered by OpenAI models and APIs
  • Completely independent — no affiliation or endorsement from OpenAI
  • Closed-source kernel for safety and determinism
  • All outputs carry provenance by design
  • Human-in-the-loop on every significant step
  • Donation-based model keeps focus on quality over scale

Media

Qyvaria AI Studio
Qyvaria AI Model

Global Radio beta

A lightweight, client-side radio app backed by the public Radio Browser API. Search worldwide stations, play instantly, and keep personal favorites saved in your browser. No sign-in, no server-side storage.

Some stations may not play due to codec or CORS. Try others if one fails.
Now Playing
—
Favorites
Add a custom stream

Qyvaria™ Policy, Terms, and Licensing

Effective Date: March 17, 2026
Owner: John Havlasek

1. Overview

Qyvaria™ is a human-operated AI studio and orchestration system developed and operated by John Havlasek.

Qyvaria may utilize services and models provided by OpenAI as part of its workflow.

Qyvaria is an independent project. Unless expressly stated in writing, Qyvaria is not affiliated with, endorsed by, certified by, or sponsored by OpenAI.

2. Licensing Structure

Qyvaria operates under a dual-layer licensing model:

  • Proprietary Core Software License (Closed Source)
  • Permissive Output License (User Deliverables with Attribution Requirement)

Each is governed separately.

3. Qyvaria Software License (Proprietary Core)

3.1 Ownership
All components of Qyvaria (the “Core System”) are proprietary and exclusively owned by John Havlasek. This includes, but is not limited to:

  • Qyvaria.py (kernel)
  • Internal agents and modules
  • Orchestration systems and workflows
  • Prompts, routing logic, and architecture
  • Internal tools, datasets, and methods
  • Branding and identity assets

3.2 License Grant
No rights are granted to any person or entity to copy, modify, distribute, sublicense, reverse engineer, decompile, or commercially exploit the Core System, in whole or in part, without prior written permission.

3.3 Restrictions
You may not replicate or reproduce system architecture, extract internal logic or workflows, redistribute any part of Qyvaria, or use Qyvaria to build competing systems.

3.4 No Open-Source Grant
Qyvaria is not open-source software. No rights are granted under any open-source license, including MIT, GPL, Apache, or similar.

3.5 Reservation of Rights
All rights not expressly granted are reserved.

3.6 Enforcement
Unauthorized use may result in legal action for intellectual property infringement.

3.7 Disclaimer
The Core System is provided “as is” without warranties of any kind.

4. Qyvaria Output License 1.0

This license applies only to outputs generated for users, not to the Core System.

4.1 Scope
Covers: Code, Text, Documents, Images and media, Datasets.
Excludes: Qyvaria software, Internal systems, Proprietary methods.

4.2 License Grant
You are granted a worldwide, non-exclusive, royalty-free license to use, modify, adapt, publish, distribute, and commercialize the outputs generated for you.

4.3 Conditions
You must:

  • a. Provide Attribution (Required)
    Minimum: “Created with Qyvaria™ using OpenAI services.”
    Preferred: “Created with Qyvaria™ (Qyvaria AI Studio), using OpenAI services. Qyvaria kernel and orchestration remain proprietary. OpenAI is not affiliated with or endorsing this work.”
  • b. Preserve Provenance
    Maintain metadata, logs, attribution markers, or embedded credit when reasonably possible.
  • c. Indicate Modifications
    Clearly state if outputs are materially altered before redistribution.
  • d. No False Endorsement
    Do not imply sponsorship or endorsement by Qyvaria or OpenAI.
  • e. Third-Party Compliance
    You are responsible for complying with applicable third-party rights and licenses.

4.4 Ownership
Qyvaria does not claim ownership of user-specific outputs solely due to their generation. This license does not transfer ownership of Qyvaria core software or Qyvaria branding.

4.5 Termination
This license terminates upon material breach. Reinstatement may be granted if resolved.

4.6 Disclaimer
Outputs are provided “as is” without warranties.

4.7 Limitation of Liability
Qyvaria is not liable for indirect, incidental, or consequential damages arising from output use.

5. Inputs and Responsibility

You are solely responsible for your inputs, prompts, and uploaded data as well as legal compliance of usage.

6. Acceptable Use

Prohibited uses include: illegal activity, privacy violations, harassment or abuse, malicious software or unauthorized access, harmful or unsafe activities.

Qyvaria reserves the right to refuse or terminate service for violations.

7. Intellectual Property & Trademarks

Qyvaria™ branding is protected. No rights are granted to use logos, brand identity, or trade dress except for truthful attribution.

8. Relationship to OpenAI

Use of OpenAI services is governed by their own terms. Qyvaria does not modify or override those terms.

9. No Professional Advice

Outputs do not constitute legal, medical, financial, or professional advice.

10. Changes to Terms

These terms may be updated at any time. Continued use constitutes acceptance.

11. Governing Structure

In case of conflict:
• Qyvaria Software License governs the Core System
• Qyvaria Output License governs outputs

12. Contact

Qyvaria1ai@gmail.com

Privacy Policy & GDPR Compliance

Effective Date: April 01, 2026
Data Controller: John Havlasek (Qyvaria AI Studio)

1. Introduction

Qyvaria is a one-person, donation-based AI studio operated by John Havlasek. We respect your privacy and are committed to protecting your personal data in accordance with the EU General Data Protection Regulation (GDPR).

2. Data We Collect

  • Email address when you contact us or send a brief
  • Donation information (processed via Ko-fi or direct payment)
  • Prompts, inputs, files or instructions you voluntarily provide
  • Basic technical data (IP address, browser info) for website functionality

3. How We Use Your Data

To respond to your inquiries, deliver services, process donations, maintain provenance records, and comply with legal obligations.

4. Third-Party Processors

Your inputs may be processed by OpenAI (our sub-processor). OpenAI’s privacy policy applies: OpenAI Privacy Policy.

Donations are handled by Ko-fi.

5. Your Rights (GDPR)

You have the right to access, rectify, erase, restrict, or object to the processing of your personal data. Contact us at Qyvaria1ai@gmail.com to exercise these rights.

6. Data Retention & Security

We keep data only as long as necessary. Reasonable security measures are in place, though no system is 100% secure.

7. Contact

Qyvaria1ai@gmail.com

Work with Qyvaria

Ready when you are. Send a brief — you’ll receive a plan, risks, and a small scoped first step with receipts.

Contact
Qyvaria1ai@gmail.com
Use Qyvaria ↗ Join the Forum ↗ Support Qyvaria ❤️ Send a Brief ✉️