Sector · healthcare
AI in healthcare: more time for care
For healthcare organizations, their support departments and consultancy in healthcare. AI takes over admin work: minutes, quality reports, roster puzzles and questions about protocols. Without medical decisions, with a human who always approves, data in the EU and a way of working in line with NEN 7510.
- Working to NEN 7510
- Data stays in the EU
- No medical decisions
- InIncident reports and figuresfrom the organization's systems
- AIAI organizes and summarizestrends, deviations, recurring themes
- AIDraft in your formatquality report per location
- HumanQuality advisor reviewsinterprets, adds and signs off
- DoneReport to the board
Data in the EU · working to NEN 7510
Worked with us
01 / What it is
What can AI do in healthcare, and what can't it?
In healthcare, AI can take over a lot of admin work: drafting minutes and reports, preparing quality reports, puzzling out rosters and answering questions from protocols and policy. With us, AI makes no medical decisions, makes no diagnoses and gives no treatment advice. A healthcare professional or employee always checks what AI prepares.
That admin is no small problem. According to the Berenschot survey (2025), healthcare professionals in long-term care spend on average 34 percent of their working time on admin, while they consider 23 percent acceptable. The target from the Integrated Care Agreement (Integraal Zorgakkoord) and a letter to parliament from the Dutch Ministry of Health, Welfare and Sport (VWS, 2025) is at most 20 percent by 2030. You won't close that gap with one measure, but every set of minutes and every report that gets done faster helps.
This page is for healthcare organizations, their support departments (quality, HR, planning, operations) and consultancy firms in healthcare. Where the work touches client or patient data, extra requirements apply. We arrange those before a single line of code is written.
01We do
Minutes, reports, rosters, knowledge base, policy documents, questions from employees.
02We don't
Diagnoses, treatment advice, triage or other medical decisions.
03Always
A human who approves, data in the EU, a way of working in line with NEN 7510 and the GDPR.
- of working time in long-term care goes to adminBerenschot (2025)
- 34%
- is what healthcare professionals themselves consider acceptableBerenschot (2025)
- 23%
- maximum admin time the Integrated Care Agreement aims for in 2030Integraal Zorgakkoord; VWS letter to parliament (2025)
- 20%
- is where the data stays, on your own infrastructure or with a European providerNinA AI way of working
- EU
02 / Sound familiar?
The care is done. The report isn't.
Most admin work in healthcare is necessary. But a large part of it is retyping, searching and summarizing. That's where AI can help, without taking anything away from the care itself.
The minutes
“After every meeting, another hour writing up the minutes. Which nobody reads again until something goes wrong.”
The quarterly report
“Every three months, a week lost to gathering figures from four systems.”
The roster
“Three people off sick, two on holiday and a collective labour agreement. The planner is still juggling in the evening.”
The protocol
“It's in the quality handbook. Somewhere. In version 4 or version 6.”
The new colleague
“Onboarding takes weeks, because all the knowledge sits with the permanent staff.”
03 / What we automate
How does AI reduce the admin burden in healthcare?
By taking over the groundwork for tasks that take a lot of time and have little to do with care. These are processes in healthcare that are good places to start. In all of them, a human approves before anything is saved or sent.
Record-keeping
Minutes of meetings
From a recording or notes to draft minutes with decisions and action points. The chair checks.
Quality
Quality reports
Incident reports and figures organized, trends and deviations flagged, a draft in your format.
Planning
Rosters
A draft roster that takes availability, leave and the rules into account. The planner decides and publishes.
Knowledge
Protocols and policy
An assistant that answers questions from your own quality handbook, always with the source and the version.
HR
Questions from employees
Leave, training, expense claims, the collective labour agreement. Answers from your own policies, personal questions go straight to HR.
Onboarding
Onboarding new colleagues
A go-to source for the first weeks, so permanent staff don't have to explain the same thing ten times a day.
Operations
Accountability and policy documents
First drafts of annual reports, policy plans and accountability documents from earlier work.
Consultancy in healthcare
Analyses and advisory reports
For healthcare consultants: data and interviews organized, a report outline in your own format. The advisor finishes it.
04 / Human and AI
Which role does AI get for which task?
In the APO method, we choose one of four roles for each task: Human, AI assists, Human approves or Fully AI. In healthcare, 'Fully AI' almost always stays empty, and that's exactly right.
| Role | Why | |
|---|---|---|
| Minutes of a team meeting | Human approves | AI prepares the decisions and action points, the chair checks. |
| Quality report | AI assists | AI organizes and flags, the quality advisor interprets and is responsible. |
| Draft roster | Human approves | AI works through the rules, the planner knows the people and decides. |
| Question about a protocol | AI assists | An answer with source and version. If in doubt, it goes to the quality advisor. |
| Treatment decision or diagnosis | Human | AI stays out. That's for the healthcare professional, and we don't build it. |
| Conversation with a client or family | Human | Attention, trust and presence. This is where a human belongs. |
Examples; we determine the role per task together with you in the AI Quick Scan.
05 / Set up safely
NEN 7510, the GDPR and a human in the loop.
In healthcare, trust is the foundation. How big the risk of standalone chatbots is became clear in August 2024: the Dutch Data Protection Authority (Autoriteit Persoonsgegevens) received, among other things, a report about an employee of a GP practice who had entered patients' medical data into an AI chatbot. That's why we arrange the agreements on data, access and checks before we build.
Working to NEN 7510
NEN 7510 is the Dutch standard for information security in healthcare. The revised version appeared in December 2024 and aligns with ISO 27001 (Kiwa, 2024). We work to NEN 7510 and ISO 27001; certification is in preparation.
Data stays in the EU
On the organization's own infrastructure or on our servers with a European provider. Azure OpenAI is an option.
In line with the GDPR
Health data is special category personal data and has extra protection. Purpose, legal basis, data processing agreements and retention periods are settled before anything goes live.
Always a human in the loop
Nothing goes into a record or out the door unseen. The employee or healthcare professional checks and approves.
Access per role
An assistant only sees what the user is allowed to see. Everything is logged, so you can always trace who did what.
No lock-in
Code, data and documentation belong to the organization, and can move within 30 days. You can always carry on independently.
06 / Where we stop
Why we don't build AI for diagnosis or treatment.
Because that's a different field, with different rules. Under rule 11 of the European Medical Device Regulation (MDR), software that provides information for decisions about diagnosis or treatment usually falls into class IIa or higher. According to the European guidance MDCG 2025-6 (2025), such an AI system is then generally also high risk under the AI Act, with its own process for assessment, documentation and supervision.
There are specialized parties for that. We focus on the work around it: the admin, the reports, the planning and the knowledge that keep healthcare running. That's where the admin burden is, and that's where AI can already help today without touching the care itself.
If a question that heads towards medical decision-making does come up in a conversation, we say so honestly and refer you on.
01No diagnosis or triage
Our AI doesn't assess complaints, symptoms or test results.
02No treatment advice
What happens with a client or patient is up to the healthcare professional.
03But the work around it
Admin, reporting, planning and knowledge. That's where the burden is.
07 / Record-keeping
Can AI help with record-keeping in healthcare?
Yes, as a draft that a human checks. For meetings without client data, such as a team meeting, a policy meeting or a board meeting, that's straightforward: from a recording or notes to minutes with decisions and action points, which the chair reviews.
When it comes to reporting on clients or patients, stricter requirements apply. Then we only build within the organization's own environment and agreements, with access per role, logging and a healthcare professional who always checks and approves the report personally. What goes into the record remains that professional's responsibility.
And sometimes the answer is: not yet. If the agreements on data, or the system it needs to land in, aren't in order yet, we start with the processes without client data. There's often enough to gain there already to get started.
01Start without client data
Team meetings, quality, HR and operations. Quick wins, low risk.
02Then, with agreements
Client reporting only within the organization's own environment, with access per role.
03The professional signs off
What goes into a record is always checked and approved by a human.
08 / The baseline
Where does your organization stand with AI?
In the AI Quick Scan, we measure AI maturity on seven dimensions: strategy, data, technology, people, processes, governance and culture. In healthcare, governance carries a lot of weight: who decides what happens with data, and who checks what AI does? Below is an example, not a client figure.
Profile · now vs. target
2.7
now
4.1
target after 12 months
Biggest jumpData readiness (2 to 4)
What we measure, on a scale of 1 to 5
Strategy and Vision+1 jump
How clear and shared is the AI vision at board level? Is the strategy written down and who holds the mandate?
now 3, target 4.Data readiness+2 jump
Is data centrally available, structured and accessible for automation and AI?
now 2, target 4.Technology and Infrastructure+1 jump
Which tools, cloud platforms and integrations are in place? How easily do you connect new systems?
now 3, target 4.People and Skills+2 jump
Does the team know AI tools, prompting and automation? Is there capacity and willingness to learn?
now 2, target 4.Processes and Adoption+1 jump
Are processes clearly described and actively optimised? How did earlier change programmes go?
now 3, target 4.Governance and Risk+2 jump
Is there policy for AI use, data ethics, privacy and security? How is risk weighed against speed?
now 2, target 4.Culture and Change+1 jump
How does the organisation deal with change? Is there room to experiment, fail and learn?
now 4, target 5.
09 / How to approach it
From first conversation to a working assistant.
We build within the AI Partnership. It starts with the € 6,000 AI Quick Scan: one day on site, and within a week a business case per opportunity and a roadmap.
- 01
Intro call
Fifteen minutes
Where is most of your admin burden, and which agreements on data are already in place?
- 02
AI Quick Scan
One day plus a week
Conversations with the board, support staff and frontline staff, processes mapped out, a business case and a roadmap.
- 03
Agreements up front
First weeks
Data, access, data processing agreements and the role of the human per task. Together with your privacy and security officers.
- 04
First process live
Quarter 1
Usually a process without client data. A human approves everything, and every month you see what it delivers.
10 / Questions
Frequently asked questions about AI in healthcare.
What can AI do in healthcare?
In healthcare, AI can take over admin work: drafting minutes of meetings, preparing quality reports, making draft rosters, answering questions from protocols and policy, and helping new colleagues get up to speed. NinA AI doesn't build AI for diagnoses or treatment advice, and a human always checks what AI prepares.
How does AI reduce the admin burden in healthcare?
By taking over the groundwork: retyping, searching, organizing and summarizing. According to Berenschot (2025), healthcare professionals in long-term care spend on average 34 percent of their time on admin. AI makes a draft, the employee checks and approves. That way the time goes to checking instead of typing.
Does NinA AI work to NEN 7510?
Yes. We work to NEN 7510 and ISO 27001; certification is in preparation. In practice that means: data stays in the EU, access per role, logging, documented data processing agreements and always a human in the loop. Where needed, we build on the healthcare organization's own infrastructure.
Do you build AI for diagnosis or treatment advice?
No. Under the European rules for medical devices (MDR, rule 11), software that provides information for diagnosis or treatment decisions usually falls into class IIa or higher, and is then generally also high risk under the AI Act (MDCG 2025-6). That's a field of its own. We focus on admin, reporting, planning and knowledge around care.
Can healthcare staff use ChatGPT for reports?
Not with client or patient data in a standalone chatbot. In August 2024, the Dutch Data Protection Authority received, among other things, a report from a GP practice where an employee had entered patients' medical data into an AI chatbot. Work with an environment in which the data agreements are settled, access per role is arranged and the data stays in the EU.
Where does the data stay when we work with NinA AI?
In the EU: on the organization's own infrastructure or on our servers with a European provider. Azure OpenAI is an option. An assistant only sees what the user is allowed to see, everything is logged, and code, data and documentation remain the organization's.
Do you also work for consultancy firms in healthcare?
Yes. For healthcare consultants, we can for example build assistants that organize data and interviews and make a first outline of an advisory report in the firm's own format. The advisor remains responsible for the analysis and the advice.
What does AI cost for a healthcare organization?
You start with the € 6,000 AI Quick Scan. After that, the AI Partnership runs on package S (€ 4,500 per month, 30 hours), M (€ 7,500 per month, 50 hours) or L (€ 12,000 per month, 80 hours), for a minimum of twelve months. We only start if the business case shows the partnership earning back at least twice the investment in the first year. All amounts excluding VAT.
More time for care, less for admin?
Book an intro call. Together we'll look at where the admin burden is, and which agreements on data are needed first.
01
Intro call
Fifteen minutes, no strings. Where does work get stuck and what do you want to achieve?
02
AI Quick Scan
One day on site. Within a week, a business case and a roadmap.
03
AI Partnership
A dedicated team that builds, trains and shows what it delivers every month.
160+ organizations · Sessions rated 9.3 · Based in Amsterdam