How we reached 88% daily AI use in four months and why it matters for sustainability

Now that AI can do so much, where do we add the most value? 

That was the question we asked ourselves earlier this year. Generative AI was everywhere, and our team was already experimenting with it. But every consultant had their own tools, their own prompts and their own way of working. Nothing was shared, and nothing could scale. 

So we started with what our clients value most: specialist expertise, clear advice and the personal touch. That’s where we want our advisors’ time to go. Everything else, we set out to automate as much as we could, so we can deliver even more for our clients. 

Four months later, 88% of our 45-person team uses AI every day. Dust, the AI platform we built our approach on, has published the full story as a case study. Here are the highlights. 

How we did it 

Rules first, agents second 

Before building anything, we set up a cross-functional AI Task Force with at least one person from every department. Together, they wrote our AI Ethical Use Policy, built on five principles: accountability, transparency, privacy and confidentiality, accuracy and reliability, and environmental responsibility. Every output has a named person responsible for it. 

The Task Force then mapped our workflows on two axes: how much value they create and how feasible they are to automate. That gave us a roadmap. It also made clear which decisions should always stay with people. 

Why Dust 

We hold data for more than 400 clients, so security came first. Dust reads from the systems we already use and copies their access rights, so nothing is stored in Dust itself. It also lets us use different AI models for different tasks. As our co-CEO Felix Keser puts it,  

“We strongly believe there won’t be one winning LLM.”  

And it works for every team, from consulting to sales, marketing and HR, under one policy and one shared agent library. 

From ten builders to 100+ agents 

Once ten people started building agents, we had ten different ways of doing it. So we built an agent that builds agents. We added a register of every agent and what it’s used for, weekly AI office hours, and a human review step on anything a client sees. With help from Dust partner Elements Agents, we later audited the whole library and trimmed what wasn’t earning its place. 

The results after four months 

Our biggest lesson? Spend most of your time defining the problem. In Felix’s words, 

“80% of your time defining the problem and 20% building the solution.” 

What this means for our clients 

Clients receive the same work from the same people, who stay accountable for it. What’s different is where our advisors’ hours go. They spend less time chasing documents and coordinating handovers, and more time on analysis, advice and conversations with clients. Before a client call, AI makes sure nothing slips through the cracks, even when a team is running six projects at once. 

We think this is where sustainability consultancy is heading. The work is getting more complex, from CSRD and EcoVadis to climate risk and decarbonisation. Clients need advisors with time to think. Next, we’re exploring how to make our deliverables more interactive, so clients can explore their results in one place instead of across separate reports and spreadsheets. 

Want the full story? Read the case study on Dust, including the policy, the prioritisation matrix and the lessons we learned along the way: How Nexio Projects reached 88% daily AI usage in 4 months with Dust