blog-self-host-breakeven1200 × 675 px
When does self-hosting an LLM beat API pricing?
A worked model for finding your break-even point in tokens per month, including the costs teams usually forget.
Blog
Field notes on private AI.
Practical writing on GPU economics, model deployment and AI security, from the engineers doing the work.
blog-quantization1200 × 675 px
Quantization without regret: AWQ, GPTQ and FP8 compared
What you lose, what you save, and how to test it on your own tasks before you ship.
blog-rag-permissions1200 × 675 px
Retrieval that respects permissions
Why most RAG demos leak data across teams, and the patterns that stop it.
blog-gpu-utilization1200 × 675 px
Your GPUs are probably 30% busy
How to measure real utilization, and the five fixes that recover the most capacity.
blog-eval-harness1200 × 675 px
Build your eval set before you pick a model
A practical guide to evaluation suites made from your own users' questions.
blog-ai-security1200 × 675 px
The OWASP LLM Top 10, in plain language
Each risk explained for engineering leaders, with the control that addresses it.
Next step
Get new articles by email.
One article a month, no marketing fluff. Email us to join the list.