[ PROOF OF CRAFT // SYSTEMS & PROTOTYPES ] HONEST STATUS LABELS • ZERO FABRICATED METRICS

Active Systems, Internal Diagnostics & Creative Pipelines.

Rather than publishing anonymous or exaggerated case studies, Built by Kellz documents the actual systems we engineer and operate—complete with explicit status labels (PROTOTYPE, IN DEVELOPMENT // RESEARCH, PRODUCTION WORKFLOW) and clear capability boundaries.

STATUS: ACTIVE DEVELOPMENT // PROTOTYPE .NET 10 MAUI + HYBRID INFERENCE
Official GOLEM G-Sigil

GOLEM — Graph-Optimized Execution Mind

Configurable Local & Hybrid AI Desktop Architecture

GOLEM is Built by Kellz's flagship local-first AI architecture—designed to unify graph-structured Markdown memory (23-directory Memory Vault), multi-agent orchestration, and hardware-honest inference across everything from a low-memory laptop to a multi-GPU workstation.

Instead of forcing users into a single model runtime, GOLEM implements an 8-Mode Universal Inference Matrix: an API-Only Thin-Client Mode (0 MB local model RAM) for light laptops and phones, plus local adapters for llama.cpp (llama-server / llama-swap), Ollama, LM Studio / Jan / KoboldCpp, ONNX DirectML / NPU, remote Kaggle Dual-T4 GPU tunnels, and ML-engineer controls for vLLM / SGLang / TensorRT-LLM.

CAPABILITY & STATUS LEDGER TRUTH VERIFIED
What Is Verified Today: Deduplicated 23-folder Memory Vault (`00_SYSTEM`–`22_TAGS`, `99_ARCHIVE`), 8-mode Universal Inference Matrix specification, Cerebral Cockpit telemetry design, and `builtbykellz.com` artifact hooks.
What Remains In Prototype Build: The `.NET 10 MAUI` desktop shell is under active development and is not yet sold as a packaged consumer binary download.
How Clients Benefit Now: Businesses can commission our Tier A.3 Custom AI Systems service to deploy grounded RAG, local/hybrid inference, or structured Obsidian/Markdown knowledge vaults tailored to their team.
STATUS: IN DEVELOPMENT // RESEARCH ARCHITECTURE NON-ADVISORY // SIMULATION & TELEMETRY

Project Domination — Multi-Agent Market & Crypto Intelligence Architecture

Regime Classification • Deterministic Risk Gates • Paper-Trade Simulation • Human Approval Boundary

Project Domination is an internal R&D systems project exploring how specialized software agents can ingest heterogeneous market data—order-book liquidity snapshots, on-chain flow metrics, macroeconomic calendar events, and technical regime indicators—and synthesize structured research briefings without emotional bias.

The engineering focus is on failure-mode containment: separating probabilistic signal synthesis from deterministic risk-limit checks, enforcing paper-trade validation gates, and requiring explicit human operator review before any action occurs.

1. Data & Regime Layer

Normalizes exchange WebSocket feeds, volatility regimes, and timestamped news sources with explicit staleness detection.

2. Deterministic Risk Gate

Hard-coded position sizing, drawdown circuit breakers, and exposure caps that an LLM agent cannot override.

MANDATORY FINANCIAL & NON-ADVISORY DISCLOSURE

Project Domination is an internal software engineering and research project. It is not a public investment product, managed account, copy-trading service, or signal group.

Nothing on builtbykellz.com constitutes financial, investment, legal, or tax advice. Built by Kellz does not publish fabricated win rates, backtest guarantees, or promised returns. Digital assets and financial markets involve substantial risk of loss.

STATUS: PRODUCTION WORKFLOW // CREATIVE R&D UNREAL ENGINE 5 • REAL-TIME BROADCAST • ASSET AUTOMATION

UE5 Virtual Production & Automated Media Pipelines

Real-Time 3D Environments • Broadcast Overlays • Structured Post-Production Workflows

Beyond web and workflow engineering, Built by Kellz develops real-time 3D environments and virtual production pipelines using Unreal Engine 5 (UE5)—powering our live technical broadcasts, gothic-cyber visual staging, and educational walkthroughs on the Watch hub.

We combine deterministic asset-naming and scene-assembly scripts with real-time lighting (Lumen), high-detail geometry (Nanite), and automated transcription/chapter indexing so live streams convert cleanly into searchable on-demand technical replays.

Applied Capabilities

• Custom 3D broadcast stage design & real-time scene switching.

• Automated VOD packaging, transcript generation, and VideoObject schema publishing.

• Custom visual direction & brand worldbuilding for technical creators and product launches.

STATUS: INTERNAL PROTOTYPE // SERVICE ENGINE MULTI-ENGINE CRAWL • SCHEMA PARITY • INDEXNOW

Search Intelligence System (Working Internal Name)

Diagnostic Crawler & Entity Parity Checker Powering Our Search Visibility Baseline

To deliver consistent, evidence-based Search Visibility & AI Answer Readiness audits, we built an internal diagnostic pipeline—currently referred to by its working name, the Search Intelligence System.

It inspects target URLs for HTTP status integrity, canonical/robots conflicts, initial-HTML content visibility, JSON-LD structured data parity against visible DOM copy, and IndexNow endpoint readiness across Google, Bing, DuckDuckGo, and Brave.

Truthful Product Status Notice (Spec v5.0 Section 3): The Search Intelligence System is an internal prototype and diagnostic workbench used by Kelly Greene during client engagements. It is not sold as a finished self-serve SaaS subscription. Clients access its capabilities today through our scoped Search Visibility Baseline service.
What the Prototype Verifies
  • • Crawl & Canonical Ledger: Detects duplicate URLs, broken internal links, and mismatched canonical tags.
  • • Schema vs. Visible DOM Parity: Flags JSON-LD fields that do not appear in visible page text.
  • • AI-Answer Readiness Check: Evaluates entity definitions, scope clarity, and FAQ structure for generative citation accuracy.