HARMONDEG Lightweight Upstream AI Safety  harmondegtech.com

System Safety & Operational Stability Technologies

Protecting Systems. Preserving Stability.

HARMONDEG develops lightweight technologies to protect system operations and identify emerging risks. Our current engineering focus is on advanced AI and AGI safety frameworks, modular protection, and controlled intervention.

Why HARMONDEG

Why HARMONDEG

Local-First Operation

Designed to run locally without requiring protected customer content to be sent to third-party servers.

Lightweight & Easy to Deploy

Efficient, modular protection designed for straightforward deployment alongside existing infrastructure.

Lower Operating Costs

Designed to reduce computing, storage, energy use, and ongoing operational overhead.

Privacy & Customer Control

Privacy-conscious monitoring with customer-defined authority over data and protective actions.

Traceable Accountability

Records relevant security events and response decisions for verification and accountability.

Low Maintenance Overhead

Manageable updates, modular replacement, and removal when no longer required.

Explore Our Work

Explore Our Work

01

Frameworks and SDKs

Modular safety architectures and specialized technical components.

Frameworks and SDKs →
02

Applications

Standalone software products for installation and practical use.

Applications →
03

Insights

Research, engineering perspectives, technical cases, and validation findings.

Insights →
Work with us

Technical Evaluation · Integration · Licensing · Partnerships

Page 3 of 7

Frameworks and SDKs

HAS-Core

Modular AI Safety Architecture

Specialized components for detection, monitoring, protection, and controlled response. Use one module or combine several.

Status labels: Research Concept Prototype Internal Validation Preview Released

Technology Areas

Data integrity and anomalous input

Privacy and information exposure

Runtime behavior and system boundaries

Safety orchestration and response

Module Directory

This directory introduces research modules and technical components. A listing does not imply that every module is currently released as a public SDK or licensed product.

SDIR

Synthetic-data integrity.

SOD

Anomaly detection.

PAC

Poisoning signals.

CPC

Adversarial-input signals.

PID

Privacy protection.

DAEF

Data-exfiltration signals.

Drift

Drift monitoring.

DRD

Duplication detection.

Subject Guard

Runtime boundaries.

Ctrl

Authorization and response.

Deployment

  • Modular evaluation
  • Supplementary protection
  • Customer-defined policies
  • Technical assessment first
Page 4 of 7

Applications

Standalone applications

Standalone Security Applications

Development and validation

HEMIG Mini Guard

Lightweight Endpoint Safety Application

Lightweight endpoint safety: observes selected signals, detects anomalies, and responds within its supported scope.

Personal terminalsAnomaly detectionProtective responses
Page 5 of 7

Insights

Research and cases

Research, Perspectives and Cases

Technical Insights

Perspectives on upstream risk detection, lightweight monitoring, privacy-first security, dynamic system behavior, runtime protection, and modular safety engineering.

Research and Analysis

Selected notes on AI risk mechanisms, systems-level safety questions, operational anomalies, technical trade-offs, and emerging directions in security architecture.

Cases and Validation

Technical demonstrations, internal test summaries, and experimental scenarios. Each case identifies its context, test conditions, scope of observation, and applicable limitations.

Selected Topics

  • Why early detection matters.
  • Protection without unnecessary content inspection.
  • Abnormal behavior in changing runtime conditions.
  • Modular safety in existing environments.
  • Evidence standards for testing.

Research background: harmondeg.org

Page 2 of 7

About

Company

About HARMONDEG

Harmondeg Lightweight Upstream AI Safety Ltd.

System Safety Technologies | Operational Stability | Advanced AI & AGI Protection

Who We Are

HARMONDEG is an independent technology research and development company based in Calgary, Alberta, Canada, specializing in system safety, operational stability, and advanced protection technologies.

Our current development focuses on the safety challenges of advanced artificial intelligence and emerging artificial general intelligence (AGI), particularly the identification of emerging risks, abnormal system behavior, operational vulnerabilities, and the mechanisms through which local disturbances may develop into wider system failures.

Through our HAS-Core technology initiative and independent safety applications, we are developing lightweight, modular, privacy-first protection technologies designed to complement existing infrastructure and strengthen the safety of increasingly complex systems.

Our long-term research direction extends beyond artificial intelligence to the broader challenges of maintaining safety, stability, and reliable operation across interconnected technological systems.

Our Mission

Our mission is to advance practical system safety technologies that identify risks closer to their origins, support timely and proportionate intervention, and help systems maintain reliable operation under changing conditions.

We aim to make effective system protection more accessible through technologies that reduce unnecessary computational demands, infrastructure investment, operational costs, and maintenance complexity.

We believe safety technologies should strengthen existing systems without imposing excessive burdens on the organizations and individuals that depend on them.

Our Research Foundation

HARMONDEG's technology development is informed by Harmondeg Philosophy, an original ontological framework concerned with the fundamental conditions of existence, universal interconnection, mutual responsiveness, dynamic stability, and coexistence through mutual development.

This philosophical foundation guides our approach to identifying root causes, understanding interactions among system components, and examining the conditions under which risks emerge, propagate, and affect overall system stability.

Our research emphasizes five interconnected principles:

Root-Cause Inquiry

Understanding the underlying conditions and mechanisms that generate risks, rather than focusing exclusively on their visible consequences.

Universal Interconnection and Mutual Responsiveness

Recognizing that system components operate through continuing relationships and interactions, where changes in one area can influence the behavior of others.

Dynamic Stability

Examining how systems maintain essential functions, respond to disturbances, and recover under continuously changing conditions.

Coexistence and Mutual Development

Supporting technological development that respects human agency, operational responsibility, and the sustainable functioning of interconnected systems.

Practical Transformation

Translating foundational understanding into engineering methods, protective mechanisms, testable technologies, and practical applications.

These principles provide an intellectual direction for our work. Engineering performance is established through implementation, experimentation, and technical validation.

Further philosophical research is available through the Harmondeg Institute for Philosophy & Practice: https://harmondeg.org/

Our Technology

HAS-Core: Modular System Safety Architecture

HAS-Core is our principal technology development initiative, bringing together specialized safety detection, monitoring, protection, and control components.

The architecture addresses risk areas associated with data integrity, anomalous inputs, adversarial behavior, privacy exposure, information leakage, operational drift, runtime boundaries, and protective intervention.

HAS-Core is designed as a modular system that can support different implementation requirements, including independent component deployment, supplementary protection layers, and integrated safety configurations.

Our development approach emphasizes adaptability, operational efficiency, privacy protection, and compatibility with existing systems.

Standalone Safety Applications

Alongside our integrated safety architecture, HARMONDEG develops independent software applications designed for direct installation and practical use.

HEMIG Mini Guard is one of our development initiatives, focusing on lightweight endpoint safety monitoring, abnormal behavior detection, and controlled protective responses.

Standalone applications are being developed with emphasis on local operation, resource efficiency, straightforward deployment, and user control.

Individual products progress through development, technical validation, and release stages before commercial distribution.

Our Core Advantages

100% User Autonomy: Full Ownership and Control

HARMONDEG is designed around a fundamental principle: users retain complete autonomy over their safety systems, data, operations, and protective decisions.

Our local-first architecture enables customers to deploy, operate, configure, and manage protection technologies within their own environments, without depending on third-party cloud services, external algorithms, or remote processing infrastructure for core safety functions.

Protected data remains under customer control. Security monitoring, risk assessment, and protective responses are designed to operate locally, without requiring customer content to be transmitted to external servers.

Customers retain authority over security policies, intervention decisions, system configuration, updates, and component removal. Relevant security events and response records remain available for customer-controlled review and accountability.

The principle is simple: Your System. Your Data. Your Decisions. Your Control.

Local-First Operation and Privacy Protection

Our technologies prioritize local execution, allowing safety monitoring and analysis to operate within the customer's own environment without requiring protected data or content to be transferred to third-party servers.

This approach supports confidentiality, data control, and reduced dependence on external processing infrastructure.

Lightweight Architecture and Easy Deployment

Our protection technologies are designed to minimize computational overhead and infrastructure requirements.

Modular components support straightforward installation, integration with existing systems, and flexible deployment according to operational needs.

Lower Operating and Infrastructure Costs

By reducing reliance on resource-intensive processing and extensive external infrastructure, our engineering approach aims to lower computing, storage, energy, and ongoing operational costs.

Controlled and Traceable Protection

Our safety approach emphasizes proportionate intervention and customer-controlled decisions.

Relevant risk events, protective actions, authorization decisions, and recovery processes can be recorded to support verification, accountability, and operational review, without requiring indiscriminate recording of normal user activity or protected content.

Low Maintenance Overhead

Lightweight, modular designs support manageable updates, component replacement, and removal when required, reducing unnecessary system complexity and maintenance demands.

Early Risk Identification

Our upstream-oriented research seeks to identify meaningful abnormal signals closer to the conditions from which risks emerge, supporting earlier detection and more effective protective responses.

Local-First Safety and Accountability

Our technologies prioritize local execution, allowing safety monitoring and analysis to operate within the customer's own environment without requiring protected data or content to be transferred to third-party servers.

This approach supports confidentiality, data control, and reduced dependence on external processing infrastructure.

Our safety approach emphasizes proportionate intervention and customer-controlled decisions.

Relevant risk events, protective actions, authorization decisions, and recovery processes can be recorded to support verification, accountability, and operational review, without requiring indiscriminate recording of normal user activity or protected content.

Research, Engineering, and Validation

HARMONDEG combines foundational systems thinking with independent technical research, software engineering, experimental evaluation, and iterative development.

We investigate safety risks not only as isolated technical events, but also as outcomes of interactions among data, software components, automated processes, operational environments, and human decisions.

Our development activities include architectural design, specialized safety modules, experimental prototypes, independent applications, and technical validation.

We distinguish research objectives, experimental findings, internally validated capabilities, and commercially released products.

This approach enables us to develop new safety technologies while maintaining clarity about their current maturity, operational requirements, and demonstrated performance.

Our Vision

Our vision is to establish HARMONDEG as an innovative developer of lightweight system safety technologies capable of supporting increasingly complex and interconnected technological environments.

Beginning with advanced AI and AGI safety challenges, we aim to extend our research and engineering capabilities toward broader applications in system protection, operational stability, and responsible technological development.

We seek to make effective safety protection more practical, affordable, adaptable, and accessible while preserving privacy, human control, and the long-term reliability of the systems on which society increasingly depends.

Work With Us

HARMONDEG welcomes inquiries from technology companies, enterprise users, system integrators, research organizations, and potential strategic partners.

Areas of interest include technical evaluation, pilot projects, technology integration, licensing, joint development, and strategic investment partnerships.

Harmondeg Lightweight Upstream AI Safety Ltd.

Calgary, Alberta, Canada

Website: https://harmondegtech.com

Email: contact@harmondegtech.com

At a Glance

Location
Calgary, Alberta, Canada
Website
harmondegtech.com
Email
contact@harmondegtech.com
Research background
harmondeg.org
Founder
Hu Yang Qingyun (Qingyun Hu-Yang)
Page 6 of 7

Partnerships

Work with us

Partnerships

Choose a type below, then contact us.

1

Technical Evaluation and Pilots

Defined test scope and environment.

2

Licensing and Integration

Component access and integration.

3

Joint Development

Collaboration on a specified need.

4

Product and Distribution

Delivery to end users.

5

Investment and Strategic Partnerships

Technology direction and growth.

Page 7 of 7

Contact

Get in touch

Contact HARMONDEG

Email us, or use the form.

Contact details

Company
Harmondeg Lightweight Upstream AI Safety Ltd.
Location
Calgary, Alberta, Canada
Website
harmondegtech.com
Email
contact@harmondegtech.com

Warning: Do not include passwords, source code, or confidential data in this form.