AI CODING

Software for problems that used to be too small.

AI has changed the economics of custom software. What used to take months of development can now be tried out in days, so it suddenly pays off to solve the everyday problems of a small business, not just the big ones.

01

Automate the repetitive

The task that steals hours every week: reports, lists, reminders. That can often be automated for a fraction of what it used to cost.

02

Test the idea cheaply

Not sure something will help? I build a simple version and try it in your operation. If it doesn't help, you haven't lost much.

03

Proven in real operations

Quality isn't promised on paper, it's proven in use. What doesn't hold up in a busy day doesn't stay.

04

From a small tool to a full product

It can stop at a small tool, or grow into something bigger. HOSPO started exactly like that: one everyday problem at a time.

My principles

01

AI makes mistakes, constantly

That's not a bug, it's the nature of AI: a probability model that guesses, not a knowledge base. My job is to turn that probabilistic nature into deterministic products, and that takes time and patience. Put simply: never fully trust the AI. It doesn't know it's lying when it lies, which is my explanation of what hallucination is.

02

Understand the problem first

I build the solution in my head before I write the prompt. If I can't express what the AI is supposed to build, it's nearly impossible to know what comes out. Garbage in, garbage out applies here too.

03

Test the limits

Meet every idea with a blank page, and don't be stopped by «AI can't build that». Check first, and you'll usually be surprised by what a language model can do. But it's not magic: never trust the output blindly.

04

Stay curious, keep up

Read up, listen to people who know more than you, and think security all the time. Don't let AI take full control of the task: stay in the loop yourself. And keep moving, things are changing incredibly fast.

AI coding — Madsen Eventyr og Data — Hospo