How Stippt puts AI to work on the most sensitive administration there is
Stippt takes over the payroll, HR and financial administration of healthcare organisations, so every file holds data about employees and clients in care. They decided to use AI anyway, with the rules written down first. Today Aimable sorts their incoming questions, helps their specialists in Excel, and is the foundation their own tools are built on.
At a glance.
Customer
Stippt, payroll, HR and financial administration for healthcare organisations.
The data
Salaries, contracts, invoices and care budgets of their customers’ employees and clients.
Teams
The administration specialists in finance, payroll, HR and healthcare administration, and the technology team building Stippt’s own tools.
Use cases
Sorting 500 to 600 incoming questions a week, Excel analysis on AFAS exports, and Stippt’s own applications on the Aimable API.
Rules first, then the technology
Stippt takes over the administration and works as an extension of their customers’ teams, which means they handle the data of people in care and the people who care for them. Their customers have a data protection officer looking over their shoulder, and so does Stippt.
Most administration offices in that position do not start with AI at all. Stippt did, but only after writing down five rules: the AI never sees who a record is about, data stays in Europe, a person takes every decision with consequences, every AI action can be looked up afterwards, and the agreements are in a processing agreement with their customers, including that their data is never used to train a model.
“We wrote down the rules before we switched on the technology.”
Protection before processing
Every request passes through Aimable before it reaches an AI model. Names, client numbers and other personal details are replaced with neutral labels, the model works with the labels, and the real values are restored in the answer. The same happens with confidential business data. Specialists see it happen in front of them, and they keep the responsibility for checking the result.
The protection sits in Agentic OS, so Stippt can switch AI models without changing anything about how their data is handled. Processing stays within the EU.
“You see the sensitive data disappear step by step, and that feels good.”
Three ways Stippt uses it.
Sorting incoming questions
Stippt’s customers send 500 to 600 questions and reports a week: when an invoice gets paid, a bicycle plan, a municipality granting a client extra hours of care. Each one lands in the ticket system and used to be read and categorised by hand. Now Aimable classifies every ticket as it comes in and routes it to the right specialist, every working day, with the personal details protected first. During the test period the AI picked the right category more often than a colleague did.
Analysing Excel exports
Specialists work on AFAS exports in Excel all day, and that is exactly the data they were never allowed to paste into a public AI tool. With Aimable inside Excel they ask questions about the open workbook, build summaries and work with formulas, under the same protection as the rest of the platform. Stippt’s own step-by-step procedures, written as skills, run on the open sheet.
Their own tools on the API
Stippt was the first Aimable customer to start from the API rather than from the chat. Their technology partner ForgeWorks connects AFAS and the ticket system to Agentic OS, so a specialist opening a ticket already has the relevant information next to it, and builds skills for recurring payroll calculations that run inside Stippt’s own tools.
Built together, within weeks
Stippt started as a design partner in May 2026 with a concrete first process: the ticket classification that already ran on a self-built setup, moved onto a platform made for this. That took a few days. Our own engineers then worked through the next steps with Stippt’s team: connected servers so their own systems can be reached from a chat, skills without code for their payroll procedures, the Excel add-in, and skills that run on the open workbook because that is how Stippt wanted to use them.
What Stippt gets out of it.
Hundreds of questions a week, sorted
Every incoming question categorised before a specialist opens it, so less manual sorting and fewer tickets in the wrong queue.
Excel analysis that is allowed
Specialists analyse their own exports with AI, something they formally could not do before.
One answer for their customers
When a healthcare organisation asks what happens to its data, Stippt can point at the rules, the logbook and the platform instead of a best-effort setup of their own.
Read Stippt’s own account.
Published by Tijn van Vlerken on 20 October 2026, the original Dutch article explains Stippt’s five rules, why they chose to use AI at all, and what their specialists say about working with it.
What would this look like in your team?
Every one of these stories started with one process, found in a first conversation. Thirty minutes with the Aimable team is enough to see where yours would start.
Shall we look for your starting point?