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
Cutting LLM Costs by 85%: 5 Hidden Quality Risks to Avoid
Aggressive LLM cost optimization can silently destroy product quality. Learn the 5 hidden risks of model switching and how to cut costs without flying blind.
Claude Opus 4.6 Fast Mode: The New Frontier for Production AI
Claude Opus 4.6 Fast Mode shifts LLM deployment from model selection to inference configuration. A deep dive on latency, cost tradeoffs, and routing logic for engineering leaders.
The AI Cost Trap: Why Falling Token Prices Won't Save Your Budget
Token prices dropped 92%, yet enterprise AI spend exploded 16x. Discover why the Jevons Paradox and agentic workflows are inflating your budget and how to fix it.
The 5 Biggest Engineering Problems with GDPR-Compliant AI-test
Legal policies don't prevent data leaks. Discover the 5 biggest engineering challenges in GDPR-compliant AI—from PII blind spots to deletion—and the architectures to fix them.
AI Infrastructure Costs 2026: A Build vs. Buy Decision Guide
Stop optimizing blindly. Learn the true TCO of enterprise AI in 2026. We break down costs for vector DBs, tokens, and observability to help you avoid the Danger Zone.
How to Reduce LLM Evaluation Costs by 90% (Without Losing Quality)
Stop running exhaustive evaluations. Discover the three-tier monitoring strategy that delivers 95% of the insight for just 5% of the cost.
From €115 to €43,000: Preventing LLM Cost Catastrophes
A single AI agent caused a €43,000 bill in 4 weeks. Learn the 5 behavioral failure modes driving runaway LLM costs and the guardrails to stop them.
FinOps for AI: How to Track & Reduce LLM Costs Per Feature
Spending over €5K/month on LLMs? Learn why per-feature cost tracking is critical for AI FinOps, EU compliance, and cutting token waste by up to 50%.
The 95% Accuracy Trap: Why Multi-Step AI Agents Fail
A 95% per-step accuracy means your 10-step AI agent fails 40% of the time. Discover the math behind cascading errors and how to fix agent reliability.
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