I often get asked where Lucanet stands in its AI transformation. Less often, how we prepared the organization for it, which, for me, is the real question. Many companies have rolled out AI tools by now, but fewer can say their teams work differently because of them.
We treated AI literacy as an organizational capability, embedded it into how we operate, and moved fast enough that there was no middle ground. Now, 98% of our nearly 1,000 people use AI in their daily work, and 91% say that they’re actively supported in adapting to AI-driven ways of working.
After nine editions, we’ve covered a lot about the intelligence we're building into the CFO Solution Platform. Here I want to examine what it took to build that intelligence internally.
Access is the easy part
In my experience, when workflows, incentives, skills, and leadership behaviors stay the same, teams' preferred tools often do too. Access alone doesn’t change how work gets done.
That's the lesson that informed how we approached this shift at Lucanet. We asked ourselves: how do we become AI-native in how we operate, use it in the right ways, and hold it to account?
When generative AI arrived in mid-2023, we were early explorers. Within three months our tech team ran their first hackathon to find out what it could do and what we could build with it. We established a clear but lightweight policy framework first: guidance on which tools to use, what data could be used, and where human review was required. After that we opened access to everyone, not just technical roles, and gave people protected time to experiment with no deliverable attached.
By 2024, the foundations were in place for us to expand the surface area, adding new tools and supporting teams to run their own pilots. The hackathons built social proof: when employees saw their colleagues solving real problems with these new tools, they wanted to try for themselves.
However, while access was broader and adoption was growing, usage remained uneven. We needed to foster the organizational conditions to cement that usage into daily work.
Embedding AI as an organizational capability
We knew that voluntary adoption would only take us so far. At some point it plateaus, and the gap between the teams who are genuinely integrating AI and the ones who are experimenting at the edges starts to matter. The step change for us came when we made AI literacy and AI-supported ways of working baseline expectations, not optional extras. We embedded these expectations into performance reviews, job specifications, and interview scorecards. AI stopped being something people could choose to explore and became part of how our organization operates.
Just as importantly, we realized that structural embedding goes beyond the tools themselves. We had to look at the supporting systems: process design, curated knowledge, accountability, and clear human review. That's what separates durable organizational capability from a collection of experiments.
This is evident in two of our internally developed agents that are still in use today.
Information security and cybersecurity agent
What began as a simple chatbot evolved over 18 months into an agentic RFP tool, now known as LuCy. A lot of the effort was concentrated on the knowledge base that sits behind it, built across security, legal, and product domains and designed to be reusable across other agents. The latest version reads full Excel questionnaires, matches questions against the knowledge base, drafts answers, and flags them by confidence level for human review. In 2026, LuCy has automatically processed 360 files and answered nearly 25,000 questions across Product, Cyber, and Legal. What used to take analysts days now takes hours, which has significantly accelerated the pre-sales cycle at negligible API cost. The knowledge base is the asset. The agent is what makes it operational.
People Operations service desk
SPARK, our People Ops service desk agent, reflects the same principle in a different context. The People leadership team had already identified a clear pattern: too much team capacity was being absorbed by repetitive employee requests and transactional service work that added limited strategic value. Close partnership with IT and AI Engineering turned these insights into SPARK. In its first three months, SPARK handled more than 1,800 conversations, reduced HR tickets per employee by 17%, and achieved a post-chat satisfaction score of 4.25 out of 5. The result is faster, more consistent support for employees, while the People team gains more capacity to focus on the complex employee and business topics that need human judgment.
These examples look different on the surface, but they point to the same pattern: AI creates value when it's built into the fabric of how work happens. At that point, you move past experimentation and gain a genuine improvement in organizational efficiency.
What transformation actually demands
I want to be honest here, because this is the part that tends to get edited out of transformation stories.
Transformation is messy. There's no clean arc where you build the conditions, then scale the tools, then everyone adapts. The reality is that tools, conditions, expectations, and capabilities all move at the same time, imperfectly, under pressure, with people who are trying to do their jobs while also being asked to change how they do them.
When we shifted to agentic development and embedded AI literacy into performance expectations, our internal pulse check showed a temporary dip in company confidence. I wasn't surprised. Real change creates real uncertainty, and anyone who tells you otherwise is either not moving fast enough or not paying attention. The question is not how to avoid that dip, but rather how you move through it.
The choice we made was speed and honesty over a long adjustment period. We communicated directly and quickly, even before we had every answer settled. Because ambiguity is more corrosive than an imperfect but honest message, simple as that. When people don't know what to expect, they fill the gap with the worst outcome. Clarity, even partial clarity, closes that gap faster than waiting until the picture is complete.
That meant saying things out loud that were uncomfortable. That the shift to AI-native ways of working was real. That it would change what skills matter and what roles look like. That we were moving at speed because the alternative, a slow, careful adjustment period, would have caused us to fail and miss our moment.
What got us through was consistency and accountability on all levels. Adaptability is not just an individual skill – it's a leadership responsibility. You can ask people to adapt while staying who they are, but only if you're doing the same thing yourself: shifting course without losing what makes you credible, holding both pace and honesty at the same time.
I've been at Lucanet for nine years, and this is one of the most significant things we've done as an organization. Because really embedding AI, making it part of how we think, how we hire, how we measure ourselves and how we serve our customers, required the organization to grow in real ways. That growth wasn't painless. But it was necessary.
The 98% adoption figure is real – measured through our pulse check on frequency of use, use cases, and whether managers actively support teams in adapting to AI-driven ways of working. The efficiency that figure reflects is just as real. But what I'm proudest of is that we built an organization that moved through uncertainty with enough honesty and conviction to come out with more capability than it started with.
The same standard, applied internally
There's one more thing worth naming. Lucanet can hold itself to this standard, and move with conviction rather than urgency alone, because AI is not a departure from who we are. We've been automating work in the office of the CFO for over 25 years. That history gave us a foundation to build from, as a software company but also as an organization making the same demands of itself that it makes of its product. Embedding AI deeply into how we work is a continuation of something we've always believed: that the office of the CFO deserves better tools, and that the people building those tools should operate by the same standard.
The alignment between what we build and how we work is what makes the commitment real. Embedding AI as an organizational capability isn't a project with a finish line. It's an ongoing choice to move at speed, improve continuously, and lead with honesty through the parts of change that don't resolve cleanly.
See the results for yourself
Lucanet's workflow agents are built to close, plan, and report on your terms, all grounded in the same trust architecture we've described throughout this series. From financial modeling to emission factor mapping, our family of agents bring trustworthy intelligence into your everyday workflows.
