Sourcing PhysicsSteel, sourcing and AI, in the field
Notes
0
Updated
2026-10-21
ENTR

Why industrial sourcing works the way it does, and how to do it better with AI.

A working notebook from someone who buys steel, structures and equipment for large industrial projects. Real problems, tested ideas, and the mistakes along the way.

Read the notesHow claims are labeled

Unit cost against lead time for three suppliers Illustrative chart. Unit cost falls as lead time grows. Three suppliers sit at different points on the curve, and the cheapest one is also the slowest. Lead time Unit cost A: fast, expensive B: the compromise C: cheap, late
Illustrative curve, not data. Real ones need your own quotes.

The notebook, by maturity

Full register

Seedling

Early and likely to change

Nothing here yet.

Growing

Taking shape

Nothing here yet.

Evergreen

I would stand behind it today

Nothing here yet.

What you will find

Field cases

Procurement problems from EPC and steel projects, written as Problem, Constraints, Alternatives, Decision and Lessons. Companies are never named.

AI lab

Experiments with agents and automation on real buying tasks. Plans are published before results, and failures are listed first.

Market notes

Short, sourced notes on suppliers, trade rules and costs in Türkiye, China, Central Asia, Europe, the US and the Middle East.

Every piece says how it knows what it knows

Lived experience
I was there. Details are anonymized.
Verified
Checked against named sources, listed with the date checked.
Assumption
A working hypothesis, not yet tested.
Illustrative example
An invented scenario. Numbers are not real.