PATENTS

# AIngle — Protected Technologies & Patent Notice

Copyright 2019-2026 Apilium Technologies OÜ. All rights reserved.

This file describes the proprietary technologies, algorithms, and
architectural innovations embodied in AIngle that are protected under
applicable intellectual property laws. This notice supplements (but
does not replace) the Apache License 2.0 patent grant in Section 3.

## Protected Innovations

### 1. Semantic DAG Architecture

A content-addressed directed acyclic graph that combines cryptographic
integrity (blake3 hashing), RDF triple storage with SPO/POS/OSP indexing,
and provenance-tracked knowledge management in a single unified data
structure.

### 2. AIngle Cortex

A cryptographic semantic layer that integrates:
- Native RDF triple store with SPARQL 1.1 query execution
- Pattern-based graph queries with namespace isolation
- Proof-of-Logic validation engine
- Zero-knowledge proof generation and verification
- REST/WebSocket API for external integration

### 3. Proof-of-Logic Engine

An inference engine implementing both forward chaining (data-driven)
and backward chaining (goal-driven) reasoning over RDF knowledge graphs,
with ontological consistency validation and rule-based inference.

### 4. Kaneru (Unified Multi-Agent Execution System)

A multi-agent framework for autonomous goal decomposition featuring:
- Hierarchical task planning with optimistic execution
- Dynamic goal tree construction and pruning
- Inter-agent coordination via semantic message passing
- Capability-based agent selection and delegation

### 5. Ineru

A neural-inspired persistence system implementing:
- Attention-weighted short-term memory with configurable decay
- Semantic long-term memory with automatic consolidation
- Importance scoring for STM-to-LTM promotion
- Content-addressed storage with semantic similarity search

### 6. Zero-Knowledge Proof Integration

Integration of zero-knowledge proof systems into a semantic knowledge
graph, enabling verifiable computation over private data:
- Pedersen commitment schemes for value hiding
- Schnorr signature proofs for identity verification
- Bulletproofs range proofs for numerical assertions
- Proof composition for complex multi-step verification

### 7. Nested Learning

A meta-optimization framework that enables adaptive model behavior
through hierarchical learning loops operating on the knowledge graph.

### 8. AIngle Minimal

A lightweight edge runtime designed for IoT and resource-constrained
devices, implementing a subset of the AIngle protocol with:
- Minimal memory footprint graph operations
- Device capability sharing via semantic descriptors
- Peer-to-peer sync with gossip protocol

## Patent Status

Apilium Technologies OÜ reserves all rights to seek patent protection
for the innovations described above in any jurisdiction. The Apache
License 2.0 patent grant (Section 3) applies only to contributions
as defined in that license and does not constitute a waiver of patent
rights for the system-level architectural innovations described herein.

For patent licensing inquiries: legal@apilium.com
For partnership opportunities: partners@apilium.com