Quick summary:
- Zensai’s “Human Success Score” borrows from Strava’s race-prediction logic to give HR leaders forward-looking workforce data instead of backward-looking exit reports
- Non-technical employees built a production-ready customer lifecycle tool in two weeks using natural language, then a three-person team used AI agents to lift renewal rates 10 percentage points
- The CEO, CTO, and CFO need to operate as a single unit on AI strategy, a framework Zensai tested internally and now tours with Microsoft as a European case study
Zensai's engineering team needed a customer lifecycle tool. The engineers had the technical ability but didn't understand the business logic. The customer experience team understood the logic but couldn't write code. So engineering handed them Lovable, a no-code AI tool, and said: Build it yourself.
Two weeks later, a working platform was in production.
That story, shared by Zensai CEO Rasmus Holst on The Disruption Is Now with host Greg Matusky at HumanX 2026, captures something happening across organizations right now. The people who understand the problem can now build the solution. The barrier between knowing what needs to happen and making it happen is collapsing. And the implications stretch far beyond HR.
Holst returned to the show to talk about Zensai's AI-powered HR platform, a new book called "The Human Success Playbook," and the management playbook he's tested internally — one that turned a three-person team into a renewal machine and accidentally produced a product Microsoft now tours around Europe.
Watch now:
Key takeaways
Language is the new code and that changes who can build
The Lovable story wasn't a one-off experiment. It was a deliberate bet that the people closest to the problem should build the solution, even if they've never written a line of code.
The CX team defined the business logic. Engineering ensured it worked at scale. The old bottleneck — waiting for developers to understand and prioritize your request — disappeared.
Holst sees this as a permanent shift, not a workaround. The platform sits inside Microsoft 365, and employees can now operate the entire HR system through natural language, asking who on their team needs a check-in, what their engagement looks like this week, who got the most kudos. The interface changed, but Holst compared it to the desktop-to-mobile transition. Same architecture, different way in.
"It's entirely built by non-coding people but enabled through our team," Holst said.
Three people and AI agents outperformed dedicated reps on 600 accounts
Zensai took its 600 smallest customers — the ones that typically get the least attention — and assigned a team of three people backed by AI agents to manage all of them.
Every morning, agents briefed the team on each customer's contract status, support history, and engagement patterns. The team could respond faster, with full context, without the usual scramble of "what did we do with them last time?" The agents remembered every interaction and personalized every message.
In four months, renewal rates jumped 10 percentage points.
The counterintuitive part: Conventional wisdom says fewer humans means worse service. The opposite happened. The previous model — individual reps juggling 100 accounts each, forgetting what happened last quarter — was actually the inferior experience.
"It becomes much more contextualized than one person having 100 customers," Holst explained. The system now scales upward to larger accounts, with humans still accountable for reviewing every message before it goes out.
AI adoption works when the CEO, CTO, and CFO sit in the same room
If you stereotype them, the CTO has the fewest people skills, HR has the least tech expertise, and the CFO cares about the money. But AI strategy requires all three perspectives at once, and Holst's book, "The Human Success Playbook," is built around what he calls this "trifecta."
He tested it inside Zensai first. A VP of Finance, the CTO, and the Chief Human Success Officer formed their own trifecta and, without being asked, designed an internal AI system built on the company's full dataset. They presented it at a Microsoft conference. Danish enterprise companies in the audience asked if they could buy it.
Microsoft now tours the team around Europe as a case study.
Agents have a skill level, and supervising them makes you a better communicator
Holst sees AI agents on a maturity spectrum. Some operate like 20-year professionals. Others function like first-day interns. Both require supervision proportional to their experience, the same way a manager would oversee a team of humans at different levels.
Matusky feels similarly. He tells his team to treat agents like junior associates who are smart and overly eager. Don’t punish mistakes. Check everything. Correct the output so the agent learns. What surprised him was how that process exposed weaknesses in the humans as opposed to the machines.
He built a skill for an agent to find what he calls “blurred stories,” under-the-radar news likely to gain traction. The agent returned generic ideas until Matusky specified that it should check prediction markets like Polymarket and look for probability signals. Once he defined clear parameters, it started delivering a useful story to his desktop every morning at 6 a.m.
AI adoption stalls without a safe environment and a selfish reason to participate
Zensai has a weekly standup where anyone in the company can show up and demo something they've built with AI. The CEO demos alongside everyone else.
That detail matters because adoption culture is set from the top. If leadership treats AI as a mandate — "use this tool, hit these productivity numbers" — people comply without investing. Holst and Matusky both argued that framing AI around career development gets more buy-in than framing it around efficiency.
Holst's company saw 60% of users adopt AI functionality within the platform, using it to build development plans and prepare for performance reviews. The weekly demos created a sharing culture where a subject matter expert could prototype an agent, hand the framework to the CTO, and watch it scale into production — something nobody planned for in advance.
"I'm a subject matter expert on a lot of stuff I learned across my career," Holst said. "If I document some of that, I can create the first agent. I can't put it into production and scale it. But if I talk to my CTO, he can easily do it."
Key moments
- How Zensai’s “Human Success Score” borrows from Strava’s prediction model (1:40)
- Why building on Microsoft Copilot paid off for enterprise adoption (2:25)
- Prompting the HR platform with natural language instead of clicking through menus (3:34)
- Why peer kudos are a valuable asset (7:35)
- Why framing AI as career-building gets more buy-in than productivity talk (10:26)
- Agents on a maturity spectrum and why accountability stays human (11:36)
- The CEO-CTO-CFO trifecta for AI decision-making (14:44)
- Zensai’s internal trifecta becomes a Microsoft touring case study (19:18)
- How the customer experience team built a lifecycle tool with no code (22:00)
- Three people and AI agents lift renewal rates 10 points in four months (29:58)
- How the next generation will start with AI and work backward (33:16)
Q&A with Rasmus Holst, CEO of Zensai
Q: You compared Zensai to Strava last year. What’s changed since then?
A: All of that’s come to fruition. It’s now out in the market and operational.
Since last year, 60% of our users are now using AI functionality in some form to build their careers and understand their performance reviews.
Q: What made you write the book as a fable instead of a business manual?
A: We started out with, let’s write it like a Porter’s Five Forces or something that becomes a textbook. Halfway through, we went, nobody knows. But if we can explain to people what the endpoint should be, then everyone else can fill in the blanks.
Right now, it’s about giving everyone a guiding framework as to how to get there.
Q: How do you think about AI’s effect on junior employees entering the workforce?
A: I famously said I never needed a mobile phone, and then my first job was at Nokia when they were selling 250 million phones a year.
I think what’s going to happen is you get a generation who just naturally can use these models much faster than we are. As such, they will not need to have the certainty of an outcome when they start. Instead, they will allow the inspiration that it takes to get to a better point.
Q: How do you get employees to share what they’re building with AI?
A: We’ve had a weekly stand-up for the longest time where everyone in the company can join, and everyone who’s built anything can show up and demo. I come up and show things I’ve built as well.
I’m a subject matter expert on a lot of stuff that I learned across my career, and if I document some of that, I can create the first agent.
However, I can’t put it into production and scale it. But, if I talk to my CTO, he can easily do it. That’s how it should be — the subject matter expert writes the framework, the tech team makes it real, and suddenly you have people across the company building things nobody planned for.
