Patent Application / Amuletplus G.K.
EVΛƎ Framework Patent Patent-Pending Architecture for Pre-Execution Governance
This page provides a public overview of the patent application for an information-processing architecture that connects human and AI decision-making, action, result recording, and re-evaluation through an asymmetric dual-loop structure and three-path feedback mechanism.
Patent Application Overview
Amuletplus G.K. has filed a patent application for EVΛƎ, an information-processing architecture designed to connect human and AI decision-making with action results through a cyclical structure.
The core of this patent application is an asymmetric connection between a consciousness loop and an action loop.
The architecture separates a consciousness loop, which handles possibility generation and selection, from an action loop, which handles action initiation, execution, result recording, and future candidate generation. Action outcomes are always returned to the consciousness side and connected to emergency response, the next origin of action, or reconstruction of the candidate space.
Consciousness Loop
Consciousness Loop
Handles the origin input, generation of possibilities, selection of candidates, and recording of decision results.
Action Loop
Action Loop
Handles action initiation, action execution, recording of action results, and generation of future candidates based on those results.
Application Number
Japanese Patent Application No. 2025-160873
Application Status
Patent Pending in Japan
PCT Application in Preparation
Applicant
Amuletplus G.K.
Technical Fields
Information Processing / Decision Support / Action Control / AI Governance
Three-Path Feedback Structure
Results obtained in the action loop do not simply terminate the process. Depending on conditions such as risk level, prediction error, satisfaction, or achievement level, the result is returned to the consciousness loop through one of the following routes.
Emergency Route
Emergency Response Route
When the risk level exceeds a predetermined threshold, the action result is immediately returned to the next input for rapid re-judgment or re-control.
Learning Route A
Learning Route A
Future candidates are generated from the action result and returned to the consciousness loop as a new origin or next action intention.
Learning Route B
Learning Route B
Future candidates generated from the action result are integrated into the candidate space to reconstruct options, assumptions, or objectives.
Functional Expression of EVΛƎ
In EVΛƎ, the dual-loop and three-path feedback structures described in the patent application are organized into four major functions so that companies and users can understand and apply the architecture more easily.
E
Origin / Input
Captures the reason a decision begins, including intent, request, observation, discomfort, purpose, or constraint.
V
Possibilities / Candidates
Organizes multiple options, alternatives, future candidates, risks, and feasibility before action.
Λ
Decision / Boundary
Compares candidates and determines execution, hold, reconsideration, or escalation to human review.
Ǝ
Trace / Re-evaluation
Records the reason for judgment, execution result, approval route, and differences, then returns them to the next decision or learning cycle.
Technical Scope
EVΛƎ does not focus only on AI output. It treats the entire process from origin, candidate generation, selection, action, result recording, future candidate generation, and re-evaluation as a continuous information-processing architecture.
Asymmetric Dual Loop
Separates the consciousness loop, which handles possibility generation and selection, from the action loop, which handles action initiation, execution, and result processing.
Condition-Based Route Selection
Selects the feedback route based on conditions such as risk level, prediction error, satisfaction, and achievement level.
Future Candidate Generation
Does not merely return action results as the next input, but generates alternatives and future candidates through a function corresponding to Va.
Structured Result Recording
Stores decision results and action results as structured data so they can be used for the next cycle, post-analysis, explanation, and self-improvement.
Candidate Evaluation and Selection
Compares multiple candidates using evaluation factors such as expected value, risk, cost, and alignment to select the next action.
Cycle from Pre-Execution to Post-Execution
Treats not only pre-execution judgment, but also state changes during execution, post-execution results, and re-evaluation as part of one cyclical structure.
How EVΛƎ Differs from General AI Use
EVΛƎ is not an AI model itself. It is a model-independent structure for decision-making, action, recording, and re-evaluation that does not depend on a specific LLM or AI engine.
General AI Use
General AI Use
The focus is often placed on what is entered and what text, image, prediction, or proposal is generated as output.
EVΛƎ Architecture
EVΛƎ
EVΛƎ focuses on the origin of a decision, which candidates were compared, why an action was selected, and through which route the result was returned to the next decision.
Application to AI Governance
The dual-loop and three-path feedback structure described in this patent application can be applied to multiple control, recording, and re-evaluation functions, including pre-execution governance for AI agents.
Pre-Execution Judgment
Before AI proceeds to action, generation, proposal, transmission, or operation, EVΛƎ can check intent, context, candidates, constraints, and decision conditions.
Decision Boundary
In enterprise implementation, control categories such as execution, hold, rejection, human review, and re-evaluation can be configured.
Human Review
Conditions that should not be judged by AI alone can be defined, and decisions can be handed over to a responsible person, manager, or specialist when necessary.
Decision Trace Logging
The origin, options, evaluation factors, rationale, execution result, and re-evaluation route can be recorded as a decision chain.
Re-evaluation
A decision does not end after one judgment. It can be updated according to results, elapsed time, changing circumstances, and new information.
Model Independent
EVΛƎ is not limited to a specific LLM or AI model. It is designed to be connected externally to existing systems, AI agents, and business workflows.
Connection to Current Business Development
Amuletplus G.K. positions this patent application as the foundational architecture of EVΛƎ and is developing it into thinking support, enterprise governance, joint verification, and application development.
EVΛƎ ARMOR
Enterprise AI governance, pre-execution control, decision trace logging, and PoC verification.
EVΛƎ MIND
A thinking method that supports human thinking, selection, action, and reflection.
EVΛƎ Workshop
Workshops for visualizing organizational decision structures and designing decision boundaries.
EVΛƎ APP STUDIO
Development of industry-specific applications, AI agents, and verification environments based on the EVΛƎ structure.
Potential Application Areas
The EVΛƎ structure is not limited to a specific industry. It can be applied to a wide range of fields where decision-making, action, result recording, and re-evaluation are required.
Enterprise AI Governance
Management of AI agent authority, decision conditions, human review, and execution history.
Decision Support
Visualization of candidate comparison and decision rationale in management, finance, hiring, screening, and review processes.
Education / Children and AI
Design of AI usage conditions and re-evaluation structures according to age, purpose, and learning situation.
Robotics / Autonomous Systems
Emergency routes based on risk levels, re-control based on action results, and learning from execution outcomes.
Copyright / Use Conditions
Structures for reflecting rights information, purpose of use, and permission conditions into pre-execution decisions.
Self-Understanding / Well-being
Visualization of personal decisions and action results, returning them to the next choice or habit improvement cycle.
Public Materials and Demonstrations
The following materials provide additional context for EVΛƎ as a pre-execution governance architecture. Some demo materials are currently Japanese-first; browser translation is recommended where needed.
Official Website
Corporate and framework information from Amuletplus G.K.
Visit WebsiteConscious Loop Demo
A public demo that illustrates the EVΛƎ decision architecture and loop structure.
Open DemoIT / DevOps Governance Demo
A demonstration of pre-execution governance for IT and DevOps decision flows.
Open DemoFinancial Governance Demo
A Japanese-first demonstration that separates AI output from execution decisions and records the decision trace before action.
Open DemoScope of Public Disclosure
This page provides a public overview of the technology under patent application only within the scope appropriate for disclosure.
Detailed technical information is disclosed only on a limited basis.
This page does not disclose the full configuration of the invention, claim interpretation, evaluation logic, or implementation specifications. The following information is not publicly disclosed for business and intellectual property reasons.
- Specific evaluation logic and weighting methods
- Detailed conditions used for route selection
- System configuration and implementation procedures
- Company-specific or industry-specific decision boundary designs
- Unpublished supplementary information related to patent claims
- Technical and operational information that should remain confidential for business reasons
Further details may be provided after the execution of a non-disclosure agreement, depending on the nature of the discussion.
PoC / Research / Licensing
PoC and Joint Verification for EVΛƎ
Amuletplus G.K. welcomes inquiries regarding EVΛƎ framework verification, AI governance design, joint research, joint development, sponsorship, licensing, and industry-specific implementation.