AI lessons from our own practice.
Beware: AI prompts age faster than you think. 7 patterns that break with every model upgrade
AI prompts that work today break tomorrow, sometimes catastrophically. Research points to two opposing failure modes during model upgrades and seven patterns that make your AI investment more brittle than necessary.
Bas van Dijk 14 May 2026 4 types of AI agents: which one fits your business?
Everyone is talking about AI agents, but no one explains which ones. We currently distinguish four, each for a different problem. Choosing the wrong one costs you time and money. We explain the difference with concrete cases from Stripe, Shopify and Cursor, plus our own experience.
Bas van Dijk 27 Mar 2026 AI replaces junior work, not juniors: what IBM, AWS and Klarna teach you
IBM is tripling junior hiring. AWS calls replacing juniors with AI 'the dumbest idea ever'. What do they know that you don't yet? And how do you select for AI fluency?
Bas van Dijk 15 Feb 2026 The periodic table of AI: how to judge whether an AI system really works
The AI Periodic Table breaks AI down into seventeen building blocks. From simple prompts to multi-agent systems. We use it as an assessment framework for AI audits. This article shows how you can do the same, without a technical background.
Bas van Dijk 14 Feb 2026 From vibe coding to agentic coding: why quality makes the difference
AI now writes almost half of all new code. That is the reality. But safeguarding quality with AI-generated code requires a different approach. We explain the spectrum, from vibe coding to agentic coding, and share how we deal with it daily.
Bas van Dijk 13 Feb 2026 7 steps to becoming AI-native (without having to figure everything out)
95% of all AI pilots deliver no measurable impact. Not because AI doesn't work, but because companies start wrong. These are the 7 steps that do work, tested in our own company.
Bas van Dijk 6 Feb 2026 Vibe coding: why 45% of AI code is vulnerable
Moltbook leaked 4.75 million records because no one checked the database settings. The founder hadn't written a line of code. This is what we learn from it.
Bas van Dijk 4 Feb 2026 AI Skills Security: 12% of skills are malicious
12% of AI skills send data to servers you don't know. The real risk: you don't know where your business data goes and can't prove it. Here's how to get a grip on it.
Bas van Dijk 3 Feb 2026 Installing Ollama securely: a step-by-step plan for local LLMs
175,000 Ollama servers sit unsecured on the internet. The cause: wrong configuration for remote access. This step-by-step plan shows how to experiment locally with LLMs without exposing your server.
Bas van Dijk 30 Jan 2026 No learnings found for this filter.