Search for a command to run...
Articles tagged with #devops
TL;DR I designed 10 LLM evaluation experiments to understand which models I should use for real DevOps workloads. I only fully ran one of them: a CI diagnostics experiment comparing Haiku 4.5 vs Son
TL;DR — Three MCP servers in, you lose track of who can do what. I built MCP Fabric, an open-source control plane that puts OPA policies, human approvals, and full audit logging between your agents an
TL;DR — 88% of organizations use AI but only 6% see real business impact. The gap isn't about tools — it's about the order you build things. AI agents work better on small, complete, vertical features
TL;DR — 14-node LangGraph pipeline. 521 tests. 88.3% field test pass rate across 15 real incidents. $0 to run on a local LLM. The biggest surprise: my test design was wrong, not the tool. The 2 AM P
TL;DR — I built a RAG assistant for on-call engineers that indexes 55 runbooks. Header-split chunking hit MRR 0.9524 vs 0.8893 for fixed-size. The embedding model, LLM, and vector DB never changed. Ch
TL;DR — I was paying 18× more than necessary because every LLM call went through one frontier model. After building a 3-tier routing architecture (Architect / Workhorse / Utility), my cost per complet
