§ 001 / NEWSLETTER
Local-first AI systems for builders running open models on controlled compute.
Benchmark rows, failure logs, VRAM fit checks, and what actually worked on controlled compute. The 5090 Reports are the flagship series, not the whole niche.
Controlled compute
Model fit costs
Agent runtime notes
Lead magnet
New subscribers get local AI build notes: local inference measurements, model-fit lessons, and the next tool shipped from controlled compute.
Get the local AI lab notes
Benchmark rows, VRAM fit checks, quant choices, and what actually runs on consumer GPUs. M-F, only when there is something worth sending.
Start here
The 5090 Reports are the flagship public artifact trail. AgentGuard is still linked where runtime limits matter, but the newsletter is about open models on controlled compute first.
Read 5090 Reports