How we built an AI-first culture at Rydoo (and what actually worked)

For years, AI at Rydoo was primarily a product story. 

Like many technology companies, we focused first on integrating AI into the customer experience. Smarter automation, better extraction, stronger audit capabilities, improved categorisation. AI was embedded into the platform long before the current wave of generative AI reached the mainstream. 

But over the last 18 months, something fundamentally changed. 

AI stopped being a capability reserved for product and engineering teams. It became a universal productivity layer. Suddenly, every department could leverage AI to work differently: marketing, finance, HR, customer support, sales, legal, operations. 

That shift changes the role of leadership. 

The question is no longer "Should we use AI in our product?" The real question becomes: "How do we build an organisation where every employee can effectively leverage AI?" 

At Rydoo, I decided to approach this as a company transformation initiative, not as a side project. What follows below is an honest account of what we built, what surprised us, and what we are still figuring out. 

This detailed account aims to answer the burning question: how to implement an AI-first culture at a growing tech organisation? 

Step 1. Building an AI guild before anything else

The very first initiative we launched was the Rydoo AI Guild. 

Before defining policies, selecting tools, or launching training initiatives, I wanted to ensure we had strong representation from across the company to collectively shape our AI strategy and culture. 

We deliberately made the Guild cross-functional and international, with representatives from every major department and geography. 

The objective was simple: create a group responsible for accelerating AI adoption internally while ensuring that every team had a voice in how AI would be integrated into the organisation. 

The Guild meets every week. We identify blockers, share best practices, review experiments, discuss new opportunities, and spread knowledge across the organisation. 

Importantly, this was not positioned as a top-down "AI committee." It became a practical operating layer between leadership and execution - a place where people actively contribute to shaping how AI is used inside the company. 

We also introduced a rotation system. The Guild is composed of twelve employees, and every month two members rotate out while two new members join. This keeps the group dynamic, regularly introduces fresh perspectives, and allows more employees to actively participate over time. 

AI adoption does not scale through PowerPoint presentations. It scales through communities, visibility, and peer learning.  

Step 2. Understanding the human reaction to AI

Once the Guild was established, we wanted to clearly understand how our teams actually felt about AI? 

There is a lot of discussion in the market around fear, resistance, and job displacement. Rather than making assumptions, we launched an internal survey across the company. 

The results genuinely surprised me. 

People were not afraid. They were excited. 

77% of respondents said they were already experimenting with AI individually. The motivation to learn and utilise AI was certainly there with 56% expressing excitement over AI transformation. What they lacked was the structure. 

They wanted clarity on: 

• Which AI tools are officially approved by the business?
• What is allowed and what is not when it comes to AI use?
• What are the best practices for AI use?
• What does "good" actually look like for AI use?

That insight shaped everything we did afterwards. It was a powerful reminder that leadership sometimes overestimates resistance and underestimates appetite. 

Step 3. Removing friction through clear technology choices

Uncertainty is one of the biggest barriers to adoption inside companies. Meaning, when employees do not know which tools are approved, they hesitate. Or worse: they use random tools without any governance. 

Based on the survey results, we made our technology choices explicit. We standardised our core AI stack around Claude and Notion AI. 

At the same time, the team introduced a company-wide AI Policy - defining what employees are allowed to do, which data can be shared with AI tools, which data is strictly off-limits, and what responsible usage means at Rydoo. 

The goal was to create trust and clarity, not to restrict innovation. 

In practice, most people want to do the right thing. They just need a framework. 

The effect was visible quickly. Within weeks of publishing the policy and standardising the stack, activity in our AI Slack channels increased and new use cases were submitted by teams across the company every week. 

The conversation moved from "Can I use AI?" to "How can I use AI better?" That cultural shift matters enormously. It defines effective adoption across the organisation. 

Step 4. Creating daily AI conversations across the company

For an AI-first company to thrive, adoption cannot live only in workshops or quarterly meetings. It needs to become part of the company's daily operating rhythm. 

To encourage that, we created three dedicated Slack channels - and the community did the rest. 

• AI Help is designed to remove friction quickly. Whenever someone is blocked, struggling with a prompt, or unsure how to approach a task, they ask the community for help.
• AI Tips is where we share practical workflows, prompting techniques, and successful experiments. It acts as a daily source of cross-team inspiration.
• AI Watch focuses externally. We share new products, market innovations, emerging tools, and best practices from other companies - continuously exposing the organisation to what is happening in the broader ecosystem.

These three channels now have a combined 429 members and generate an average of 25 posts per week. 

This may seem like a small gesture, but culturally they are incredibly important because they normalise AI discussions, make experimentation visible, and create continuous peer-to-peer learning at scale. 

Step 5. Introducing the belts: measuring AI maturity like a skill

We found early on that AI adoption varies enormously between individuals. Some employees already build workflows and automations. Others are still discovering prompting basics. 

I wanted a system that could help people understand their current level, make progression visible, and create genuine motivation around learning. 

So we built an internal AI Belt System, inspired by martial arts belt progression. 

Employees progress from white belt to black belt based on their AI maturity and practical capabilities. What makes the system particularly interesting is that it is role-specific. The meaning of a "green belt" in marketing is not identical to a "green belt" in engineering or customer support. 

We evaluate employees across six core dimensions: 

1. Applied Understanding - ability to understand AI in context and adjust behaviour accordingly
2. Core Usage - frequency and effectiveness of AI in day-to-day work
3. Workflow & Builder Capability - ability to create repeatable workflows, automations, or tools
4. LLM Maturity - ability to use multiple models and select the right one for each task
5. Impact & Outcomes - measurable effect of AI on performance and productivity
6. Collaboration & Contribution - contribution to Rydoo's AI culture and adoption

To date, we have 7% of our employees who are white belt, 33% yellow, 37% green, 17% blue, 6% brown and 1% black belt. The most common gap across roles is Workflow & Builder Capability. 

The goal is continuous progression. The AI landscape evolves too quickly for anything less. 

Step 6. Making AI visible every week

Many companies treat AI transformation as a one-time announcement. But, culture only changes through repetition. 

That is why we launched "Thank God It's AI Friday" - TGAIF. 

Every Friday, we host a one-hour session dedicated entirely to AI. The format rotates constantly: live demos, prompt workshops, use case reviews, workflow building sessions, guest presentations, or open discussions. 

We have now run 12 consecutive sessions, with an average attendance of 101 employees. The most-attended session covered Claude basics including set-up, prompting, and building projects and skills. 

Keeping AI visible every single week creates organisational learning at scale. It sends a clear signal about what leadership actually prioritises.  

Step 7. Turning ideas into execution

Finally, we moved from experimentation to operationalisation. 

We mapped AI use cases across every function inside the company, then prioritised the highest-impact opportunities. From there, we organised internal hackathons and focused working sessions to accelerate delivery. 

Many organisations struggle because they stop at inspiration. But AI transformation only becomes real when workflows actually change. The companies that will win are the ones systematically integrating AI into daily execution, as well as talking about it.  

AI adoption is a leadership challenge and priority

The biggest lesson from this journey is that AI transformation is not fundamentally a technology problem. It is a leadership and change management challenge. 

Most employees are already curious. Most teams already want to improve. Most organisations already have access to powerful tools. 

The real differentiator is whether leadership creates clarity, momentum, safety, learning systems, and cultural permission to experiment with AI. 

At Rydoo, a few structural decisions we made - the Guild, the Belt System, the tooling choices, the weekly rituals - created the initial conditions. But it only became real when the organisation took ownership collectively. That handover from CEO initiative to company culture is, I think, the actual goal. 

We are still at the beginning of this journey, and we are learning every week. 

However, this is already clear: the companies that outperform in the coming years will not simply be the ones building AI products. They will be the ones building AI-native organisations.

That all starts with leadership deciding that AI adoption matters.