For you who works with AI every day
You can see that it's useful, but you struggle to find the right form. You keep hunting for the perfect prompt, but the results vary. You don't need more tools. You need a system.
Who this is for
For you who leads people who use AI
You see people experimenting, but no two do it the same way. What you need is an architecture that makes sure the AI tools across your organisation pull in the same direction.
For you who is sceptical
You're allowed to hate AI. You're allowed to criticise AI. And that's good. But if you've never used it, you can't influence the direction it takes either.
What I believe
13 principles
01
The perfect prompt doesn't exist
The prompt isn't the problem. The problem is where you start. What you need is a system where the AI already knows what it needs to know.
02
A system that works, every day
A prompt library is a cookbook you never cook from. A system is a kitchen that works. Every day.
03
This is architecture, not engineering
AI isn't just an IT project. This is information architecture: the right information, in the right place, at the right time, for a new user.
04
AI forgets, and that's a context problem
Every conversation starts from zero. The fix isn't a longer prompt, but an architecture that remembers.
05
AI amplifies, but you decide
AI can make you better at what you're good at, or amplify the mistakes you didn't know you were making. You have to know what good looks like.
06
Quality is an architecture, not luck
Workslop is polished content with no substance. Anti-workslop demands quality gates, verification and evaluation.
07
AI lies, but the responsibility is yours
The system has to be built for source-checking, verification and human quality control.
08
Steering beats cleanup
Define what you want before the AI starts. Steer as you go. Feed the learning back into the system afterwards.
09
What you don't share can't scale
When the way you use AI lives only inside each person's head, the organisation has a systems problem.
10
Model independence is a discipline
Context files, knowledge packs and routines should move with you when the tools change.
11
You're the boss, AI is the employee
Leaders set direction, delegate responsibility and check the results. They don't micromanage every sentence.
12
Sustainability is a design choice
Use heavier models where it counts, lighter ones where they're enough, and build workflows that don't waste.
13
Everyday innovation is the goal
The goal isn't for autonomous agents to replace people. The goal is to use AI to get a little better, day by day.
What I build
A system where AI makes you better, and the other way round
I build systems where the AI knows your context, respects your quality bar, can take on responsibility without micromanagement, and gets better with every task. Everything should be portable. Nothing should be locked to a single vendor.