The technical layer, in my own words.

I design agentic workflows, not single automations: multi-step chains where each step hands structured output to the next. The pattern in its simplest form, a brief is submitted, then structured and tagged, then a project is created, then the right people are notified, removes the manual shepherding that consumes most of the time between creative steps.

I treat prompts as infrastructure rather than inputs: reusable, chainable and versioned, so a workflow behaves the same way regardless of who runs it. That is what turns a personal productivity trick into something a 30-person marketing organisation can depend on.

The operating model I use is explicit about ownership: AI owns generation, structuring, scaling and pattern analysis. Humans own judgment, prioritisation, relationships and final sign-off. Making that boundary explicit is what makes teams adopt the system instead of resisting it.

I stay hands-on rather than reading about the tooling, and I judge new tools by what is substantively useful versus what is hype. That is why I can pick the shortest path between a business need and a working system instead of adopting tools reactively. The tools I actually use →

Governance is built in from the start: GDPR across SE, DK and NO, the EU AI Act and financial-services compliance, so the systems are compliant by design rather than retrofitted. How the governance model works →