We build ambient AI systems, presence infrastructure, and developer tools that operate at the boundary where technology becomes invisible — running on your hardware, under your control.
Every project under the Liminal umbrella shares a conviction: the most powerful technology is the kind that disappears into your environment. Local inference. Privacy by architecture. Zero cloud dependency.
A model-agnostic AI gateway that runs entirely on consumer hardware. ATLAS routes queries through a dual-model HYDRA architecture — fast L1 for conversational flow, deep L2 for complex reasoning — with all behavioral intelligence living in deterministic code, not prompt engineering. The model is a hot-swappable commodity. Four model swaps, zero gateway code changes.
Algorithmic trading strategy engine built on a J-pattern detection system with an 8-state finite state machine. From zero to audited in one session — 146 tests, 54 of 58 findings resolved. Currently in live validation.
Real-time status synchronization across platforms. Any status change propagates to all tenants instantly — no precedence logic, no hierarchy. Includes CalBridge for multi-calendar sync across M365 and Google ecosystems.
CLI tool for training and deploying custom wake words. CPU-only inference, Docker-first deployment, ONNX model output. Designed for edge devices that need always-on voice activation without cloud dependency.
End-to-end voice cloning and lip-sync video generation. XTTS-v2 voice synthesis paired with MuseTalk visual generation — producing natural conversational avatars from a single reference image and voice sample.
Dual-AI orchestration system for structured deliberation. Two Claude instances collaborate through a shared filesystem and JSONL convergence bus — one drives execution, one drives critical analysis. DEV and DELIBERATE modes.
Receipt capture and categorization pipeline. Extracts line items, vendor data, and totals from photos and PDFs — feeding structured expense data into personal and business accounting workflows.
We believe the next generation of intelligent systems won't live in distant data centers — they'll run on your hardware, in your space, under your control. Every tool we build prioritizes local inference, privacy by architecture, and the principle that the best technology is the kind you stop noticing.
Our engineering philosophy is simple: all behavioral intelligence lives in deterministic code. Models are commodities — hot-swappable, benchmarked, replaceable. The gateway is the product. This makes our systems vendor-proof, auditable, and built to outlast any single model generation.
If it can run on your hardware, it should. Cloud is a fallback, not a foundation.
Four model swaps, zero code changes. The architecture survives any vendor decision.
Intelligence in code, not prompts. Testable, auditable, reproducible behavior.
Your data never leaves your network. No telemetry, no training, no exceptions.
Liminal Technologies Group is the IP holding company and development lab for Gaelen — a systems architect who builds at the intersection of AI infrastructure, enterprise operations, and edge computing.
With deep roots in enterprise technology rollouts, design systems, and process automation, Gaelen brings an operator's mindset to every build. The same discipline that standardizes software across complex organizations also architects AI gateways that survive four model swaps without a code change.
Background spans enterprise ERP implementations, VBA/Python automation at scale, AI systems integration, and the kind of hands-on engineering that only comes from years of solving real problems in production environments — not labs.
Inquiries, partnerships, consulting engagements, and interesting problems welcome.
[email protected]