Minutes After Every Consultation: Bringing AI Summaries to Dutch Hospitals
Written by Andrea Rossi.
What we learned from launching our first AI feature in healthcare, and why we’re about to take it to the next level.
You finally made it to your doctor’s appointment. After weeks of waiting, this is your chance to explain what’s been bothering you. But as you talk, your doctor is typing away, and somewhere in the back of your mind, a thought creeps in: Are they even processing what I’m saying? Should I slow down so they can keep up?
You’re not imagining it. A large portion of a doctor’s time during each consultation is spent on documentation, and that time is taken directly away from the conversation with you.
At ConsultAssistent, we’re working to give that time back to doctors. That’s why, one year ago, we developed our first AI-powered feature: AI summaries of consultation reports. Before I get into what we’re building next (there’s a more in-depth article coming on that), I want to share the story of what we’ve achieved so far and what it took to roll it out responsibly in a hospital setting.
How does ConsultAssistent prepare for the consultations before the doctor arrives?
When a hospital or clinic schedules an appointment, ConsultAssistent handles patient communication: invitations, reminders, and smart questionnaires that patients fill out online, at home, before their visit to the hospital or clinic. The questions are the ones a doctor would otherwise have to ask at the start of the consultation—about symptoms and how long and how severe they have been, medication use, medical history, and so on.
I say “smart” because the questionnaire adapts as the patient answers it. This ensures that only the relevant questions are asked, while the rest are skipped. For example, if you answer “no” to shortness of breath, you won’t be asked ten follow-up questions about it. This way, the patient spends less time filling out the questionnaire, answers questions that are relevant to their situation, and the doctor gets a clearer picture, which he or she can process more quickly.
Based on those answers, we generate a structured medical report and provide it to the doctors and nurses who need it before the consultation begins. The doctor walks in already familiar with the patient’s history, can prepare in advance if anything stands out, and doesn’t have to spend time during the consultation on routine or administrative questions. Thanks to this, the appointment becomes more of a genuine conversation.
Once the report is ready, how can ConsultAssistent help save even more time?
Our reports are based on a template system that we have refined over the years in collaboration with doctors from partner hospitals and the medical professionals on our own team. Different answers provided by the patient trigger the generation of different sections of the report.
Our users are very satisfied with the quality of our reports. However, a good medical report must be comprehensive and should be concise and structured. That works well for preparing for a consultation, but not for the patient’s medical record. For that, doctors need the opposite: a brief summary consisting of a few easy-to-read paragraphs that are easy to skim.
Until recently, they typed those out by hand, either during or after every single consultation, and that time really adds up after a full day. The problem was clear: we had automated the generation of the report from the questionnaire, but compiling the summary was still a manual task.
Turning dense, structured information into a few clear paragraphs happens to be something large language models (LLMs) are really good at. We’re not asking the model to make a medical judgment here. We’re asking it to save the doctor some typing.
AI summaries save 2–4 minutes per consultation
In April, our AI summaries went live in three Dutch hospitals. The feedback from doctors has been encouraging: in 90% of the reviews we received, the summary was approved and ready for use.
The impact: Doctors save at least two to four minutes per consultation, depending on the type of consultation. If you work outside the healthcare field, two to four minutes might not sound like much, but consider that most consultations last roughly ten to twenty minutes. We gave doctors back about 20% of that time!
And this isn’t about squeezing in extra appointments on the same day. The Dutch healthcare system is already stretched thin: healthcare workers spend about 30% of their time on administrative tasks—a figure the government wants to reduce to 20% by 2030—and without additional measures, the sector is headed for a shortage of around 300,000 workers by 2035. Every minute we save on administrative tasks helps move us in that direction.
Those minutes aren’t really “saved” anyway; they just take on a different form. The doctor remains fully present during the conversation, free from the rush and the constant switching between tasks that comes with taking notes, and with the mental space to ask the right questions (and yes, sometimes these moments simply turn into a much-needed coffee break). On the other side of the table, the patient gets to look their doctor in the eye and talk about what’s really bothering them, instead of feeling like they’re just reciting symptoms into a file. They get to feel heard.
On a personal note, these days everyone can think of an example of AI being misused: features that make us think twice about our privacy, or wonder if it was really necessary to add that feature to a refrigerator. Well, as a software engineer, being able to apply AI in a way that gives people back their time and attention is very rewarding.
Making AI Hospital-Proof: Four Design Principles
Generating a summary using a large language model (LLM) is the easy part these days. The hard part is integrating AI into a hospital's workflow while ensuring high quality.
You can’t move fast and break things in healthcare—patients’ lives are at stake. And trust doesn’t get many second chances, because a doctor who sees one poor summary will think twice before relying on the next one.
But we weren’t starting from scratch: years spent working within hospital workflows, side by side with doctors, had already taught us how to tackle this. Four principles guided our design:
1. Privacy comes first—literally. It is the first stage of our pipeline. Before a report leaves our platform, a pseudonymization step removes all personal details: the model receives the medical content along with the minimal clinical context it needs, and nothing that directly identifies the patient. Everything runs on infrastructure hosted in Europe, in the Netherlands.
2. Every specialty is different. This is where the doctors’ input comes in. An ENT specialist and a breast surgeon want different aspects highlighted in a summary, so we’ve made it possible to tailor the summaries to each specialty, and we developed and refined them in collaboration with the doctors who use them. As of this writing, we cover the Breast Clinic, ENT, Cardiology, and Proctology departments, and will soon launch summaries for Gynecology and Gastroenterology.
3. Failures should be invisible. This is how we protect hospital workflows and the doctors’ trust in them. Summaries are generated asynchronously, completely decoupled from the workflows that hospitals rely on. If generation is slow, fails, or the output doesn’t pass our quality and safety checks, the doctor simply won’t see a summary yet: the system retries in the background. Anything suspicious is rejected and regenerated, because a doctor should never be the one to discover a faulty AI output.
4. Every summary is an opportunity to learn. The doctors’ reviews close the loop. Doctors can rate and provide feedback on each summary they receive, and we’ve built tools to evaluate the impact of different prompt versions. When we make a change, we don’t have to guess whether the summaries have improved—we measure it, with the people who read them every day.
What's next: How can we further improve quality?
At ConsultAssistent, we keep coming back to the same question: How can we save the precious time spent in healthcare, help doctors be more effective, and support patients throughout their journey?
AI is opening up real possibilities in healthcare, and applying it safely is where our experience comes into play: we understand the clinical workflows, the data constraints, and what doctors will and won’t trust. AI summaries were our first step, and we already know where they can be improved.
Everything we learned this past year is shaping the coming year. We’ve been building a more advanced AI pipeline that goes beyond single-shot generation, with dedicated quality control stages along the way, plus an evaluation framework to ensure that every improvement is measurable, safe, and repeatable.
But that deserves its own article, so we'll save the details for the next post. 🙂
If you'd like to follow along, you can find us on Medium and LinkedIn. And if you're curious about who we are: consultassistent.nl/en/about-us.
ConsultAssistent helps hospitals and clinics save time on patient communication and preparation, so doctors can focus on what matters most: the patient.