Unit tests passed. Every one of them. Green across the board.
And then we ran the parser against real legacy Gerber files — files from actual PCB designs, exported by real design tools used by real engineers over the last twenty years — and the success rate was 25%.
Three out of four failed.
For thirty years I have broken work into tasks. Decompose the feature into subtasks, estimate the hours, write the code, move the ticket. The unit of progress was the line of code. The measure of a good day was how much I shipped. That loop was so deeply embedded in how I worked that I did not notice it was a loop. It was just what development meant.
Then I started delegating implementation to AI, and the loop broke. Not gradually. In about a week.
The first reaction is always disbelief.
"That's not possible." Or: "That only works for trivial problems." Or the politer version: "That must be very rough code."
So here are the numbers. Not estimates. Actuals.
Everyone tells the same story about AI-assisted development. AI generates code fast, so you ship faster. Straightforward. Compelling. Wrong.
The actual productivity gain from AI does not come from generation speed. It comes from verification infrastructure that makes it safe to accept AI output at scale. The counterintuitive truth: the team that writes the most tests ships the fastest. Not despite the testing. Because of it.
The plan was to spend a weekend validating whether a complete PCB design library was actually buildable at AI velocity.
Not a prototype. Not a demo with curated inputs. Something that could consume real Gerber RS-274X files — the manufacturing format that PCB designers actually export from KiCad, Altium, Eagle — parse them completely, and produce manufacturing-ready outputs.
I have been writing software for thirty years. In that time I have sat through thousands of daily standups, hundreds of onboarding sessions, and more planning ceremonies than I care to count. Most of them existed for one reason: transferring context from people who had it to people who did not. The new developer needs to know how the deployment pipeline works. The team lead missed yesterday's discussion about the API change. The architect needs to understand why the data model looks the way it does before approving the next feature.
These are not bad reasons to meet. But they are expensive reasons. And increasingly, they are avoidable ones.
You'd think parsing a part number would be trivial.
A string of characters. Match it against a pattern. Done.
Then you actually try to do it across a real component database, and you discover how much institutional knowledge is packed into those strings — and how little of it is documented anywhere.
There's a reason temporal analytics resonates particularly strongly in Norway.
It's not just that Norwegian organisations face the same data challenges as everyone else — though they do. It's that several converging factors make Norway an unusually well-positioned market for AI-powered data infrastructure, right now.
In late 2024 and through 2025, we published a series of short horror stories about building with LLMs. Not fictional in the sense of being made up — fictional in the sense of being slightly dramatised versions of things that happen, or will happen, or already have happened to someone you know.
The format was deliberate. Security warnings written as dry checklists get skimmed. Security warnings written as campfire stories get remembered.
Here are all eight tales.
Most data systems are built to forget.
Not deliberately. It's just the natural consequence of how databases work. When you update a customer's address, the old address is gone. When you change a product price, the previous price is gone. When a process state changes, the prior state is gone.
Each of those deletions also deletes something else: the why.