№ 001 · Legal AI4 min read
Why legal teams are moving AI closer to their documents
How local retrieval lets firms work across contracts, discovery, and matter files without sending privileged material into external AI systems.
07 · FIELD NOTES
Field Notes & Technical Library
Engineering guides, architectural benchmarks, and real-world deployment notes from testing local LLMs on private hardware.
Read Note →
№ 002 · Hardware6 min read
One Spark, dual Spark, or custom RTX PRO: which path, when
When one Spark is enough, when linking two enables frontier models, and when a custom RTX PRO workstation or multi-GPU server is the honest recommendation.
№ 003 · RAG5 min read
RAG is the first local workload most teams should test
Why document search, summarization, and source-grounded Q&A are the cleanest starting points for private AI in regulated industries.
IN PREPARATION
Upcoming Engineering Notes
Currently in benchmark testing for publication:
Cloud vs. local is the wrong first question
A framework for what belongs on frontier APIs, local workstations, or private clusters.
Target Audience: Founders, operators
NVFP4, Q8, FP8, and BF16: How Precision Impacts Performance
How model size, quantization, serving engine, and context length shape user experience.
Target Audience: Technical buyers
What a private AI assessment should measure
Throughput, latency, concurrency, retrieval quality, security path, and cost crossover.
Target Audience: Teams deciding what to buy