Field notes.
Real opinions about what AI is getting right and wrong for operators in 2026, written by us, in our own voice, not a content calendar. Plus the occasional teardown of something Izzy built and what it taught us.
Articles
Stop AI Hallucinations in RevOps with Eval Datasets
AI hallucinations cost businesses billions. Learn why vibe coding in RevOps is dangerous and how to build a production-grade eval dataset in just one week.
How Do AI Agents Work? The Complete Architecture Deep-Dive
Only 12% of AI agents in production achieve high ROI. Discover the underlying architecture of AI agents, including memory, tools, and multi-agent frameworks
5 Problems With RAG Citations in Production That Will Get You Fined, Fired, or Both
Learn the 5 fatal flaws in production RAG citations, from unfixable hallucinations to EU AI Act violations, and the architectural decisions required to fix them.
5 Hidden Problems With AI Agents in Production
Gartner predicts 40% of AI agent projects will fail. Discover the 5 hidden problems killing AI agents in production and how engineering teams can fix them.
LLM Production Engineering: The 2026 Playbook for CTOs
Stop treating AI like a science fair. Discover the 5 LLM production patterns EU startup CTOs are using to control costs, quality, and EU AI Act compliance.
How to Build a Production LLM Observability Stack in 2026
A practitioner's guide to the winning stack: Tracing (Langfuse), Cost Control (LiteLLM), and Governance (PromptMetrics). Includes a Week 1 to Quarter 1 implementation plan.
Why Your LLM App Breaks at Scale: 7 Architecture Mistakes (2026)
Is your LLM bill eating your runway? Discover the 7 critical architecture mistakes killing AI startups in 2026 and the production-ready stack to fix them, from semantic caching to EU AI Act compliance.
Open Source vs. Enterprise LLM Observability: The EU CTO’s Guide
EU CTOs: Why your "free" open source LLM observability setup could cost €200K in hidden compliance expenses. A practical TCO guide for the AI Act era.
The High Cost of Silent AI Updates: Preventing $10k Weekends
From schema drift to runaway cost loops, silent model updates are a liability. Here is how defensive engineering and "Golden Sets" protect your enterprise AI.
Field notes, in your inbox
One email per week. No content calendar — just what we’re building, what broke, and what we changed our minds about.
Operator Stack
The free community where we continue these conversations between posts.
Join free