Technical Research & Engineering

GLOBAL INTENT COMPANY

Interactive technical environment for machine learning, scientific instrumentation, computational systems, experiments, and project evidence.

Developed independently by founder and researcher Cory Tortorici. Source code, controlled experiments, null results, architectural iterations, and public research logs are made inspectable across all project nodes.

Source Repositories

Laboratory Chronology

LATEST FROM THE LAB

Recent empirical findings, architectural iterations, and null results surfaced directly from active research logs.

Active laboratory research log entries with timestamped records are available in the full research log.

Interactive Spatial Knowledge Graph

EXPLORE THE RESEARCH

Navigate the connected network of research projects, architectural pipelines, and domain clusters in real-time 3D.

Rotate • Zoom • Click Nodes to Unfold

60 FPS

Tier:

Focused on KB (Kernos + BNLM) • Orbiting sub-nodes active

Domains:

SPATIAL FOCUS:KB (Kernos + BNLM)•Systems Research

GitHub Live App

Systems ResearchKind: engineeringBrowser-Native Operating Environment & Intent Router[KB-BROWSER-OE]

KB (Kernos + BNLM)

GI-Company/kb

A browser-native operating environment unifying Kernos OS systems architecture with BNLM client-side machine learning inside a single tab. Features an in-browser intent classifier trained in seconds (~8k parameters) with per-character occlusion attribution, a capability-governed VFS terminal with pipes, sandboxed Pyodide Python WebAssembly execution, and Groq cloud intelligence fallback.

Engineering Objective:Deliver a private, zero-install operating environment in the browser capable of training and running private intent classifiers entirely offline with explainable attribution, capability policies, and sandboxed Python execution.

Structural Overview

KB operates as a full client-side cognitive operating environment. Users can train a private ~8k-parameter intent router on local data in seconds, inspect decisions with per-character occlusion attribution, execute Python scripts in Pyodide WebAssembly, and navigate a capability-governed VFS terminal with zero data leaving the browser tab.

Canonical Source Artifact

GitHub Repository

Primary Source CodePublic

GI-Company /kb

Canonical public source repository: Kernos + BNLM browser-native operating environment with offline intent router and Pyodide.

Canonical GI-Company Source

LIVE DEMO VIEW SOURCE

Design Invariants & Principles

[1]Local-First Privacy: Intent classification and explainability execute entirely inside the client tab without telemetry or server round-trips.

[2]Honest Held-Out Re-evaluation: Model corrections are evaluated against frozen held-out test splits to measure authentic generalization.

[3]Capability-Based Security: Terminal commands and agents operate under explicit capability policies (can / policy built-ins).

System Components

Embedded BNLM Engine

Vendored client-side machine learning engine under src/bnlm/ providing in-browser tensor math, embedding, classification, and WebGPU compute shaders.

Intent Classifier & Occlusion Explainer

Trains lightweight local models in seconds; explain command calculates exact per-character occlusion attribution behind every decision (lib/localClassifier.ts).

VFS & Terminal Pipeline

Virtual File System with overlay support, standard shell piping, redirection, and capability policy enforcement (lib/vfs.ts, lib/terminalPipeline.ts).

Pyodide Python Worker

Sandboxed Python 3 runtime executing inside a dedicated Web Worker via WebAssembly without server dependencies (lib/pythonRuntime.ts).

Capability Security Layer

Enforces system capability policies (can / policy built-ins) governing file access, network requests, and tool execution boundaries.

Cloud Intelligence Bridge

Optional Groq API connection for synthetic dataset generation and complex queries, while intent classification stays 100% offline (lib/groqFetch.ts).

Technology Stack

React 19 & TypeScriptVite 6BNLM Sub-Engine (JavaScript / WebGPU)Pyodide (Python WASM Worker)Tailwind CSSGroq API (Optional Cloud Fallback)Vercel

Technical Decisions

Vendored BNLM sub-engine directly in src/bnlm/ to ensure reproducible in-browser tensor execution.

Pyodide isolated to dedicated Web Worker with postMessage boundary to prevent main thread blocking.

Occlusion-based attribution chosen over saliency approximations for mathematical determinism.