Local AI Workstation Decision Framework
A requirements-led way to choose local AI hardware across workload, memory, power, thermals, noise, storage, networking, privacy, budget, and upgrade path.
Read the evidence recordHardware Lab · Builds
A parts list is not evidence.
A build becomes known-good only for a stated workload, configuration, test, date, and limit. The foundation does not turn an educational example into a tested recommendation.
Local Versus Cloud Cost Inputs
A transparent input model for comparing local and cloud execution without inventing prices, collapsing privacy into dollars, or ignoring human and failure costs.
Read the evidence recordWhat 16 GB of VRAM Can and Cannot Do
A workload-first framework for reasoning about a 16 GB graphics-memory ceiling without promising that a named current model or configuration will fit.
Read the evidence record