By Becky Wright, Executive Director, Unions 21 | 5 min
We found that 46% of our management had never used AI - they had never even seen it. Our goal was to expose people to AI in a way that felt relevant to their daily work.
Maarten Leeuw, Programme Manager Data & AI, FNV
The FNV is the largest trade union federation in the Netherlands, with one million members and 1,700 staff. It faced a challenge that will feel familiar to many unions: how do you adopt AI responsibly when your organisation has very different levels of digital literacy, genuine political sensitivities about what automation means for workers, and no shared framework for making decisions?
Their answer wasn't to roll out a set of tools and hope for the best. It was to build the organisational conditions for AI to work - starting with culture, governance and genuine learning, not software.
The challenge
FNV identified three internal blockers before they started:
Fragmented experimentation: individual departments were trialling AI tools in silos, creating inconsistent approaches to data security and no shared learning.
Strategic uncertainty: real questions existed inside the union about how time saved by AI should be reinvested - for staff, or for members? This needed honest conversation, not avoidance.
A literacy gap: nearly half of the management team had never directly used an AI tool. You can't govern what you don't understand.
What they did
Rather than launching a single initiative, FNV built five interconnected strands of work. The logic was that technology adoption without culture change doesn't stick - so they invested in both simultaneously.
An AI Lab for structured experimentation
FNV created a dedicated space to collect and use cases from across the union, prioritise them by likely impact, and develop proofs of concept before wider rollout. Early tools included a social media text writer, a press invitation generator and AI-powered minutes generation to reduce administrative burden. More ambitious projects followed: a chatbot to help members navigate collective bargaining agreements, and predictive modelling for sector analysis.
The Lab approach matters because it manages risk. Experimentation happens with oversight, not in spite of it.
Community, not just communication
To prevent AI from becoming a top-down IT project, the FNV invested heavily in peer-to-peer engagement. An AI Connect Community on Microsoft Teams, combined with a regular newsletter sharing what was working (and what wasn't), created a culture of shared learning rather than managed rollout.
They also created Riven: an AI-generated mascot who 'evolves' alongside the programme. It sounds like a small thing, but making the technology feel approachable and even playful helped people move from anxiety to curiosity.
Real learning and development
FNV's Data & AI Academy has trained over 400 staff - and crucially, the curriculum isn't just technical. Alongside prompt engineering and Power BI, staff are taught critical thinking, data storytelling and ethical reasoning. The message is clear: AI literacy isn't about knowing how to use a tool. It's about knowing when to question what it tells you.
Governance that leads by example
FNV developed a formal AI policy, reviewed externally by the AI company Gartner, built around ten principles including maintaining human oversight, preventing bias and checking every outcome. Significantly, this internal governance framework is kept distinct from FNV's external advocacy on worker rights and AI regulation because unions should be able to lead by example, not just by argument.
The tenth principle, incidentally, is “have fun while learning”. That's not a throwaway. It reflects a deliberate choice to make this feel like an opportunity, not a compliance exercise.
A unified data and AI platform
The longer-term goal is a single platform where staff can query the union's data in plain language. For example, “How many members do we have in healthcare?” - and receive instant, accurate visualisations. This is about making information genuinely accessible to the people who need it, not just to those with technical skills.
What this means for your union
While the FNV is a large federation with significant resources, three things stand out as transferable:
Get your General Secretary or Director of Finance engaged early - not just for approval but as an active champion. FNV's programme had strategic credibility because a senior leader owned it. Without that, AI initiatives tend to stall at the pilot stage.
Build understanding before you build anything else. The instinct is to launch tools, but the FNV's experience suggests the more important investment is in helping people understand what AI actually is, what it can and can't do, and how to use their own judgement alongside it.
Share what you're learning with both the good and the mistakes. The AI Connect Community wasn't a showcase of successes. It was a space for honest exchange. That's what builds the trust that makes adoption possible.
Rob Kelsall, Assistant General Secretary for NAHT, the school leaders’ union recently revisited the FNV after originally visiting as part of a Unions 21 delegation.
He said:
Building capability in data and AI isn't just about technology, it's about empowering people. By equipping union teams with the tools and confidence to harness these responsibly, we can better understand member needs, anticipate challenges and strengthen our collective voice. Thanks to FNV for their generosity in sharing their experience, and to Unions 21 for making the visit possible.
