
Approach · 22 November 2026
Detecting fraud and anomalies
Fraud that was never discovered is not in your examples. How a model learns what is normal, ranks anomalies and explains every signal.
Read the article →
Approach · Healthcare · 22 November 2026 · Martijn van Grieken
Can AI make a roster that works for staff, clients and the organisation? How we approach that question, presented at R’evolutie van zorg 2026.

In many care organisations a roster is only really fixed once the shift is over. There are too few permanent staff, the flexible pool is used every week, and working hours are one of the reasons people leave. The question that then comes up: can AI help make rosters that work for the support worker, the client and the organisation?
This page describes how we approach that question, worked out in a project with organisations in disability care. Martijn van Grieken presented the approach on 19 November 2026 at the R’evolutie van zorg congress in Apeldoorn.
The support worker wants control over their own roster and to enjoy the work. The client wants support that fits their care needs, their own preferences and the location. The organisation wants a complete roster with less use of the flexible pool and of staff not on the payroll. Those interests can clash, and a good solution starts with having all three at the table.
Staff want to schedule themselves. A worked example shows why that alone is not enough when staffing is tight. Ten support workers state their preferences for four shifts that each need four people.
| Shift | Needed | Preferences | With planner |
|---|---|---|---|
| Tuesday morning | 4 | 7 | 4 |
| Wednesday evening | 4 | 5 | 4 |
| Saturday evening | 4 | 1 | 4 |
| Sunday morning | 4 | 2 | 3 |
| Total | 16 | 15 | 15 |
Illustrative calculation.
There are almost enough sign-ups, but if everyone chooses for themselves, five places remain open. A planner who distributes the preferences leaves one open and keeps most preferences intact. A Swedish pilot reports that an automated planner met around 81 per cent of nurses’ preferences, without substitute staff (Rönnberg & Larsson, 2010). The answer is therefore self-scheduling with a planner that looks at the whole, and AI can fill that role.
AI rostering is the application with the most impact, but rarely the first step. What has to be in place beforehand: clear communication about the roster, reliable data and linked systems for care, HR, finance and rostering, and management information on demand and staffing. Without that insight an AI planner has nothing to steer on. Improving communication is usually the quick win you start with. How we work on that foundation with care organisations is described under healthcare.
We work in six phases, based on CRISP-DM, the process model used for data projects since the late 1990s:
The first phase takes four to six weeks and delivers a shared problem definition, success criteria and a plan. More about the process model is described under our approach.
Bring one planning question to a first conversation. That conversation is without obligation and we aim to respond within one working day.
Author: Martijn van Grieken, founder of Twentynext. How we make this. AI wrote the first version of the Dutch original, and a person at Twentynext read and approved it before it went live. This English version was translated with AI. Twentynext is responsible for the content but cannot guarantee that it is accurate or complete. See our disclaimer.

Approach · 22 November 2026
Fraud that was never discovered is not in your examples. How a model learns what is normal, ranks anomalies and explains every signal.
Read the article →

Approach · 22 November 2026
A sold-out day hides how many customers still came. Why average sales underestimate demand, and what you know with little history.
Read the article →

Approach · 22 November 2026
Data-driven working starts with agreeing who owns which data. Eight steps, a baseline measurement across five areas and the role of the set-up lead.
Read the article →
We use Google Tag Manager to measure visits and Leadinfo to recognise which company is visiting. Neither loads unless you agree. If you choose essential only, the site works as normal and we measure nothing. Cookie statement · Privacy statement