What I have learnt from 18 months of discussions with senior management at NGOs and B Corps. An honest assessment of the current situation, and a few uncomfortable hypotheses that I’d like to put forward for discussion.
April 2026 · Noted by · reads in ~2 minutes
Eighteen months, around thirty organisations, a hundred discussions about AI in NGOs, social enterprises and B Corps across the DACH region. This is the data on which these seven theses are based. No study, no representative survey. But a pattern that has recurred, whether I was in Hamburg, Vienna or Zurich.
In my experience, organisations that launch AI as a major strategic project almost always fail. Not because AI doesn’t deserve a strategy, but because the first six months don’t need a strategy – they need experiments. Small, reversible, instructive ones.
Without exception. Whether it’s donation management, grant applications or member enquiries: the shared inbox is the natural first lever. It is the symptom of all translation problems between people and data.
A skill that the team does not trust will not be used. Trust is not built through demos. It is built through transparency. Teams need to understand what the skill does when it gets it wrong.
The BCG formula (10 per cent algorithm, 20 per cent data, 70 per cent people) is correct. I have found it to hold true in every project. Most AI projects fail not because of the technology, but because of the 70 per cent.
Small teams cannot learn twelve tools at once. The best advice I can give is: learn two tools properly. Leave the rest out.
Full automation is rarely the right goal. More often than not, the goal is a tool that empowers people to make decisions more quickly and confidently.
The most common misconception is that NGOs and social enterprises are ‘lagging behind’ when it comes to AI. My experience suggests the opposite. They have clear values, limited resources and real pressure. This makes them excellent testing grounds for the meaningful use of AI.
45 minutes, one screen shared, an honest assessment. Not a sales pitch, I show you where I see leverage and where I do not.