Braatz Research
Applied AI Research, in the Wild
Braatz Research takes state-of-the-art neural networks — Large Language Models, diffusion models, Kolmogorov–Arnold networks, vision-language and agentic systems — and puts them to work on novel business problems.
Research-grade rigor, production-ready outcomes. From cloud, fog, and edge deployment to fully air-gapped frontier-model inference in places the cloud cannot reach.
Selected Research & IP
U.S. Patent 12,324,122 B2 — Portable autonomously powered computer with thermally separated heat energy sources in vorticity-inducing cavity. A wood-chassis, fanless, undetectable AI computer designed for off-grid medical deployment.
Method and apparatus for network-enablement of devices (filed 2000) — the first formal definition of what is now called the Internet of Things, cited across the research and patent literature that followed.
Founder Jason C Braatz is a Research Fellow at Stanford and a PhD candidate at The University of Texas at Austin. Formerly VP-level AI/ML leadership at Google, Sony, and Disney; author of three hundred forty-four issued patents through that work.
What we work on
Multilayer Perceptron and Kolmogorov–Arnold hybrid architectures · differentiable simulation · quantization at the limit · RAG and agentic workflows · LLM, SLM, VLM, and VLA deployment · air-gapped frontier inference · explainable AI for regulated industries · AI governance, risk, and ethics.