Notes From Aluf
Practical notes on AI that ships.
Plain explanations of how we scope, build, and ship AI systems, written for the people who have to decide whether to trust an AI project with real budget and real stakes.
Where the Real AI ROI Is Hiding in Manufacturing and Operations
Most AI headlines are written for software companies. Most of the measurable return we have seen has been in manufacturing and industrial operations.
What “AI Agent” Actually Means, and When You Need a Human in the Loop
AI agent has become a term that means everything and nothing. This is what we actually mean by it, and when a human needs to stay in the loop.
How to Scope a Fixed-Price AI Project (and Why Most Firms Won't)
Open-ended retainers are easy to sell and easy to extend. This is how we scope fixed-price AI engagements so the price and timeline actually hold.
RAG vs Fine-Tuning: What Actually Grounds an AI System in Your Data
RAG and fine-tuning get talked about as competitors. In practice they solve different problems, and most production systems default to RAG.
Why Most AI Pilots Never Make It to Production
Most companies do not have an AI idea problem. They have a shipping problem, and it usually kills the pilot before it reaches production.