Rolling out a tool like Claude to 750-plus people takes an afternoon of admin. Building AI fluency and making it useful for everyone takes a lot more care. Ultimately, our goal isn’t adding a tool, it’s giving people time back for their craft, so they can apply the brilliant judgement we hired them for, and surprise themselves with what they’re able to achieve here.

One approach can’t fit every team
Pleo has employees spread out across Europe and beyond, all with different crafts, technical capabilities and needs. A developer, a people partner, and someone in customer experience all use AI differently, and each has different risks to watch for. Our AI enablement programme combines tailored support with central coordination to keep everything joined up.
Where support needs to be tailored to how our different teams work, we have champions representing different departments (we call these functions) at Pleo who act as a local bridge to adoption. Employees across different teams have someone to turn to who can connect general guidance to specific use cases and help with questions and blockers, because they understand the work.
Alongside our function champions, our technical teams focus on making sure the tools we use work reliably and safely, while a cross-functional team supports bringing in new Model Context Protocol (MCP) connectors. With their knowledge of the tools, laws and regulations, they help get connections up and running and ensure each connector goes through the right technical and security teams, so no team is left to handle this alone.
Our central team helps local successes become company-wide practices through the AI Hub, learning opportunities, cross-functional office hours, and reusable workflows.
Learning where people already gather
Long structured courses or a series of in-person training days don’t suit our pace or ambition to build an AI-fluent culture. We want people to learn context-specific skills, find ways to apply them straight away, and immediately notice the difference in their work.
Ongoing learning happens in moments our teams already share. People share experiments and useful patterns in their function meetings. Alongside these we host practical sessions on themes such as using AI as a second brain, codifying your work, and building an AI Chief of Staff.
Offsites and hackathons give teams a space to dream up AI solutions for problems they actually face day-to-day. At a recent offsite in Lisbon, our Enablement functions ran an AI hackathon to redesign parts of the employee journey, and within a week of getting home, a few of those ideas were already in our workflows.

Keeping judgement human
Claude is a useful tool for gathering context, preparing a first draft and improving workflows, but it can’t transform how we work on its own. We need the people behind each craft to bring the expertise and judgement that make the result good. Some examples from across Pleo:
- an engineering manager connecting Claude, Slack, and their calendar to propose changes to a busy week;
- a Claude skill that turns fragmented engineering signals into a shared radar on customer frictions;
- customer and meeting information brought together to prepare call briefs and draft follow-ups for an account owner to review;
- and a People agent that supports the comparison work for our new levelling system, while people make the final decisions.
Enabling AI with care
As a regulated business we build privacy, security, and responsible access into how we enable AI. Our legal, security, and technical teams assess use cases, review integrations and keep guidance current as the technology and regulations evolve. That means our teams are confidently using approved integrations with appropriate permissions, and can have use cases reviewed by raising a ticket in Slack.
We also want AI usage to deliver real business value, with spending that is visible and proportionate. We help businesses manage their spend, so we’re mindful of our own. Token allowances help us manage costs. We share practical guidance on getting value from the tools, and when someone needs more budget, a simple request form captures what they want to achieve, which company goals it supports and the impact they expect. That helps us unblock promising work while keeping a clear link between AI spend and business value.
What comes next
We’re still learning what works, and for whom. A recent employee survey showed us that 84% of our employees feel AI is relevant to their work, and over 90% of employees feel at least moderately confident in using it. We still have just under 50% of employees who are asking for more support. Right now, we’re seeing great signals that Pleo’ers are starting to feel confident using AI to take away repetitive or cumbersome tasks, beginning to reshape workflows and sharpen their focus on their craft. Next, our focus is to continue building on that while consciously moving from access and adoption to building AI-native workflows, measurable impact and value creation, and deeper capabilities.
What we’re hearing from our people:
“I’m AI first with a portfolio of skills in Claude Code that helps me deliver quality work consistently and at speed.” (AI Pulse survey respondent)
“AI adoption needs more than tools. It needs clear strategy, technical support, leadership and protected time to learn and experiment.” (AI Pulse survey respondent)
About the people tackling the rollout
Miriam Tocino
Chief of Staff, Offering
Responsible for driving Pleo’s company-wide AI transformation, tying together strategy, governance, tooling, and learning to make AI part of everyone at Pleo’s normal working day.
Ermis Zacharopoulos
Director, Enablement AI
Responsible for leading AI and systems change across our Enabling functions, building solutions alongside the team to improve how we work with AI and keep sensitive data secure.