From GenAI experiment to shared quality guideline - a best practice

This story didn't start at a boardroom table but on the shop floor. During a survey in the Christmas period of 2024, employees themselves indicated that they needed support in writing transmission reports. That question - straight from practice - gave the initiative a strong undercurrent from the start: a shared interest and a high dose of goodwill. Everyone was looking in the same direction.

To answer that question, the 'GEM transmission' experiment started. In October 2025 the ethical council judged that the experiment in its then form was too vaguely described and didn't sufficiently align with the spirit of privacy legislation, the legal framework and the AI Act. Instead of dropping the initiative, that moment was seized as a turning point: the experiment was reshaped into a thoughtful, project-based approach. It was precisely that reorientation that made the trajectory sharper, sturdier and ultimately more widely supported.

Every organisation knows two currents. The overcurrent is the visible and measurable: the content, the structure, the rules, the planning. The undercurrent is what lives beneath it: trust, relationships, emotions and meaning. A plan in the overcurrent only truly succeeds when the undercurrent comes along - and the other way around: a strong undercurrent only delivers lasting results when the overcurrent is also clear and supported.

By developing trust and connection alongside content, structure and planning, both currents reinforced each other and the work naturally started to flow more smoothly and faster.

Organisational developers as bridge-builders. The leads had strong ties with both the central services and the province. As 'trusted faces' they brought credit and safety with them beforehand, and they formed the indispensable link for short communication lines.

Vulnerability as strength. Instead of a top-down rollout, the team started from mutual dependence: 'We can't do this alone.' By making each other's expertise explicitly indispensable, a culture of equality emerged.

Respect for the existing order. Existing structures and hierarchy were recognised, not bypassed. By continuously involving provincial directors, managers and the sounding-board group through short lines, resistance was prevented before it could arise.

Iterative feedback loops. By constantly moving between expertise and daily practice, supported solutions emerged instead of one-sidedly imposed instructions. The micro-example of the disclaimer shows this rhythm: expert advice → test against the business → adjustment → supported solution.

The direct result was there: by bundling expertise and iteratively adjusting the plans, the end goal was achieved. But the real win sits in the system effect. A blueprint emerged for successful multidisciplinary work.

The distance between central services, provinces and shop floor got smaller. Employees now experience for themselves that a transparent feedback culture pays off, and experts see the direct added value of testing in practice.

Build a dual network. Make sure the leads of a project are anchored in multiple parts of the organisation - central and provincial. They are the oil in the machine.

Embrace the 'we-need-you' mindset. Make mutual dependence explicit. That lowers thresholds and invites real co-creation, instead of defensive behaviour.

Organise the translation. Set up a structured process in which central advice is always first tested against the business - and the other way around - before implementation.

In short: this project succeeded through integral work, with attention to overcurrent and undercurrent. Those who put connection, trust and shared meaning first let content flow naturally and build results that last.