How Personio achieved their biggest-ever drop in support contact rate during their busiest month

Personio is Europe's leading HR software platform, serving 16,000 companies and 1.5 million employees. With that scale comes a support operation that handles about 40,000 conversations a month in one of the most heavily regulated industries, where customers rely on Personio to manage sensitive employee data, payroll, and recruitment workflows. That volume spikes sharply every January, when annual HR policy resets trigger a flood of questions from employees and admins.

Their previous solution, a search tool with a conversational interface, had done its job. But it was never built for the volume, complexity, or self-serve demands of a fast-growing enterprise product; it couldn't reason or handle questions that went beyond what was explicitly written down and lacked robust mechanisms to enforce a high quality bar. They needed an AI Agent that could perform at real volume, integrate into Personio app, and be manageable day-to-day by a small in-house team.

Personio evaluated several AI support agents before landing on Fin. One of the strongest contenders relied on a forward-deployed engineering model, while Personio wanted a solution that their small team could own and manage independently, without relying on a vendor for every change. "Their philosophy was to always have an engineer that would do a lot of the stuff for us, when in reality we wanted a tool that was simple enough for us to control, configure, and maintain over time," recalls Alessandro Presa Perez, CX AI Operations Lead at Personio.

Fin's self-manageable approach won that argument. So did the partnership. “What made a big difference was having Fin's own support team sharing their knowledge – people who run their own support org with Fin every day, sharing what works and how to think about building for AI-first support. That was by far one of the biggest weaknesses in the other vendors. The other vendors never thought beyond the technology.”

Personio's deployment wasn't straightforward either. Rather than embedding a standard chat widget, they wanted Fin to live inside their own product, so when a Personio customer clicks "get support," they stay within the Personio app. Fin’s Deployment Services team worked shoulder-to-shoulder with Personio's engineers to make it happen, navigating every technical curveball along the way. "I don't know how we would have done this without Deployment Services," Alessandro says. “The value went far beyond implementation, they helped us rethink processes, ownership, and how we operate in an AI-first support model.”

Personio went live ahead of the seasonal spike. "We wanted to deploy by January because that's when we see the biggest spike in volume, and it was really important that we met that deadline," says Alessandro. “We managed to do that with Fin, and we started seeing results pretty much straight away.”

The Personio team saw fewer customers needing to reach a human agent, with Fin resolving issues that would previously have landed in the support queue. The global human support contact rate dropped 25–30% in Q1, and in the small business segment, where all customers must go through Fin before reaching a human, the drop was 40–50% year over year. Fin now resolves around 80% of all conversations, well above what the team had scoped for. "We had assumed we'd be in the 60 to 70% range," Alessandro recalls. "We frankly did not anticipate how well Fin would perform."

Beyond the numbers, the team found something they hadn't had with their previous solution: visibility. “Before Fin, we couldn’t clearly see where customer experience was breaking down. Now we understand where the problems are and why,” shares Alessando. “Fin surfaces where we’ve massively under-documented product behavior, where product limitations are generating volume, and where our internal processes need to improve”

With January behind them and Fin resolving 30,000 monthly conversations, the team is now focused on building Procedures, enabling Fin to automate more complex, multi-step workflows like permissions and access troubleshooting.