I did not set out to build a platform. I set out to stop rebuilding the same spreadsheet.
You know the ritual. The clause you need is on page 214 of a PDF. The table you need is in a book on someone else's desk. The spreadsheet that does the check was built by a colleague who left two years ago, and nobody is quite sure which cell holds the assumption that matters. So you open six tabs, re-derive a formula you have derived a hundred times before, and the actual engineering - the judgement, the part you trained for - gets whatever time is left over.
I did that one time too many.
So I built a calculator. Then another one. A handful of Eurocode checks, the ones that kept coming back, built once and built properly, so I would never have to rebuild them again. It was meant to be a private fix for a private annoyance.
Then other engineers found them.
That was the first surprise, and it quietly changed what the thing was. A tool you build for yourself is allowed to be a black box. You know what is inside it, you know what you assumed, you know exactly where not to trust it. A tool that other people run their designs through is not allowed to be a black box. It owes them an explanation.
So I made a rule, and it has held ever since: show the working. Not a green tick. The assumptions, the clause number, the intermediate values, the formula in the same form the code writes it. An answer you cannot defend in a review is not an answer. It is a liability with a tick next to it. If CivilAxis tells you a section passes, it tells you why it passes, which clause says so, and what it assumed to get there - so you can go and check it against the book, and disagree with it if you want to.
That rule is expensive. It is far slower to build a tool that shows its reasoning than one that returns a number. It is also the only reason the tools are worth anything. Before an engine ships, it gets checked against a published worked example: an SCI guide, a textbook problem, the standard's own example. If my number does not match theirs, my number is wrong, and I go and find out why. Some of the most useful days of this project have been the ones where a design guide proved me wrong.
Then came the second surprise.
I built it in English, and I assumed the readers would be English too. They were not. The engineers who kept coming back were Vietnamese, and today the Vietnamese pages out-draw the English ones several times over. So the platform became properly bilingual, and TCVN became a first-class code rather than a translation. TCVN 5574 is not Eurocode 2 in Vietnamese. It is a different code with a different philosophy, and it earned its own engine, checked against the Vietnamese textbooks engineers here actually learn from. Then came AISC. Then a searchable standards library. Then a page for every single steel section size. Then an AI that had genuinely read the standards and cites the clause instead of improvising one.
Somewhere in there it stopped being a set of calculators and became a platform.
One more thing has never changed: it stays free. Engineering knowledge should not sit behind a licence, and this profession is far bigger than the group of people who can expense a seat. Students are checking sections on it. So are consultants on a deadline. Both should get the same answer.
What I write here
The articles are the same idea in a different shape. Worked examples that run through the same verified engines behind the tools, so the numbers on the page are real numbers. Explainers on what a clause is actually asking for, and what changes when you carry a check from Eurocode into TCVN or AISC. Case studies on real structures: the ground they sit on, the one decision that saved them, and the physics that nearly went the other way. And notes on what is new here.
In English and Vietnamese, because the engineers reading it work in both.
Based in Singapore. Still building, still checking my own numbers against the book.