Practical writing, not thought leadership.
Short pieces on workflows, automation and the economics of both — written from engineering work, with arithmetic you can re-run on your own numbers. No invented case studies, ever.
Articles
Start with the problem you have.
When should you replace an Excel workflow?
Five signs the spreadsheet has quietly become a system — and what to do before it breaks.
EconomicsHow much does manual quotation actually cost?
A worked example, in ringgit, using numbers you can check against your own week.
TechnicalWhat AI should — and shouldn't — do in engineering software
Why every number in our systems comes from code, and what we let AI touch.
In the pipeline
Coming next.
Written when there is something checkable to say — not on a content calendar.
- Coming: What a Workflow Blueprint contains — the document you keep even if you never commission the build.
- Coming: When automation isn't worth it — the arithmetic behind an honest no.
- Coming: Approvals that live in WhatsApp — moving sign-off out of chat without adding bureaucracy.
- Coming: Buying industrial equipment from overseas — what a technical and commercial bridge actually checks before you pay.
House rules
Checkable, or not published.
Three rules govern everything here. First, no invented proof — no composite "client stories", no metrics we didn't measure. Second, worked examples use arithmetic you can re-run with your own inputs; if a number appears, you can trace where it came from. Third, where AI comes up, we state the rule we build by rather than a position on the discourse:
AI reads and writes. Code calculates. AI handles unstructured work — reading an enquiry, classifying a document. Deterministic code produces every number, so results are auditable and repeatable.