AI in the consulting room: how do you build trust when technology listens in?
AI in the consulting room: how do you build trust when technology listens in?
Rob Lagendijk
Rob Lagendijk
11 October 2026
An AI assistant can help document a consultation. But as soon as technology starts listening, the conversation with the patient changes as well. For healthcare organisations and healthtech companies, successful introduction therefore starts with a simple question: can people understand and assess what is happening?
An AI assistant can help document a consultation. But as soon as technology starts listening, the conversation with the patient changes as well. For healthcare organisations and healthtech companies, successful introduction therefore starts with a simple question: can people understand and assess what is happening?
Imagine a consultation in which the doctor spends less time typing and more time looking at the patient. Voice based documentation promises exactly that: support in turning a conversation into a draft clinical note. It is an attractive idea. But it also creates new questions for the patient. Is my voice stored? Who can access it? And who checks whether the record is accurate?
Imagine a consultation in which the doctor spends less time typing and more time looking at the patient. Voice based documentation promises exactly that: support in turning a conversation into a draft clinical note. It is an attractive idea. But it also creates new questions for the patient. Is my voice stored? Who can access it? And who checks whether the record is accurate?
Those questions belong to the implementation itself. If an organisation only answers them after concern appears, it is already behind.
Those questions belong to the implementation itself. If an organisation only answers them after concern appears, it is already behind.
AI use in general practice is growing. That does not yet prove time savings.
AI use in general practice is growing. That does not yet prove time savings.
In the Nivel survey of Dutch general practices covering 2025, almost half of participating practices reported using AI supported applications. Voice based reporting was used by 36 percent. The results were based on responses from more than 800 practices, so they are not a complete count of all Dutch general practices. [1]
In the Nivel survey of Dutch general practices covering 2025, almost half of participating practices reported using AI supported applications. Voice based reporting was used by 36 percent. The results were based on responses from more than 800 practices, so they are not a complete count of all Dutch general practices. [1]
Use and proven effect are different things. The Dutch College of General Practitioners describes opportunities for AI, but also stresses that scientific evidence for the effectiveness of many applications remains limited. With speech recognition, human review remains important because errors and loss of nuance are possible. [2]
Use and proven effect are different things. The Dutch College of General Practitioners describes opportunities for AI, but also stresses that scientific evidence for the effectiveness of many applications remains limited. With speech recognition, human review remains important because errors and loss of nuance are possible. [2]
The communication task is therefore concrete: explain what the application does today, which result has actually been measured and where human judgement remains necessary.
The communication task is therefore concrete: explain what the application does today, which result has actually been measured and where human judgement remains necessary.
Three conversations that should happen before launch
Three conversations that should happen before launch
1. With the healthcare professional: what changes in my work?
1. With the healthcare professional: what changes in my work?
A promise about less administration only becomes meaningful in daily practice. How much correction does a draft note require? Who reviews it? Does the application work within existing systems? And what happens when it fails?
A promise about less administration only becomes meaningful in daily practice. How much correction does a draft note require? Who reviews it? Does the application work within existing systems? And what happens when it fails?
Involve users in shaping the story. Their experience shows which benefits hold up in practice and which objections need to be solved first. A critical question is useful implementation information.
Involve users in shaping the story. Their experience shows which benefits hold up in practice and which objections need to be solved first. A critical question is useful implementation information.
2. With the patient: what happens to my conversation?
2. With the patient: what happens to my conversation?
A patient does not need to understand the technical architecture of a language model in order to ask relevant questions. Give clear, application specific information about what is recorded, where data is processed, who can access it and how long it is retained.
A patient does not need to understand the technical architecture of a language model in order to ask relevant questions. Give clear, application specific information about what is recorded, where data is processed, who can access it and how long it is retained.
Also explain how questions, objections and alternatives are handled. The explanation at reception, on the website and in the consulting room should match the actual design of the system. Reassuring copy cannot repair missing agreements.
Also explain how questions, objections and alternatives are handled. The explanation at reception, on the website and in the consulting room should match the actual design of the system. Reassuring copy cannot repair missing agreements.
3. With the board and supplier: which promise can we actually make?
3. With the board and supplier: which promise can we actually make?
More time for the patient is a substantial claim. Agree in advance how it will be tested. Measure documentation time including corrections, the quality of the record and the experience of professionals and patients.
More time for the patient is a substantial claim. Agree in advance how it will be tested. Measure documentation time including corrections, the quality of the record and the experience of professionals and patients.
Also determine who answers difficult questions and how incidents will be discussed. That creates a story that can still hold when a journalist, patient representative or employee asks for detail.
Also determine who answers difficult questions and how incidents will be discussed. That creates a story that can still hold when a journalist, patient representative or employee asks for detail.
From product promise to credible evidence
From product promise to credible evidence
For healthtech companies, this creates a positioning opportunity. A supplier that clearly explains which users an application is suitable for, how it fits their work and which results have been achieved under which conditions makes the conversation with a healthcare organisation more concrete.
For healthtech companies, this creates a positioning opportunity. A supplier that clearly explains which users an application is suitable for, how it fits their work and which results have been achieved under which conditions makes the conversation with a healthcare organisation more concrete.
A small pilot can provide useful evidence, as long as its limitations remain visible. Describe its size, duration, method and outcomes. A positive result in one practice does not justify a general claim for the whole healthcare system.
A small pilot can provide useful evidence, as long as its limitations remain visible. Describe its size, duration, method and outcomes. A positive result in one practice does not justify a general claim for the whole healthcare system.
The strategic conclusion is straightforward: build communication into implementation from the beginning. That brings promises, practical experience and expectations together earlier.
The strategic conclusion is straightforward: build communication into implementation from the beginning. That brings promises, practical experience and expectations together earlier.
Five questions for your organisation
Five questions for your organisation
Can we explain in ordinary language what the AI does? Do employees know how to assess its output? Can we answer patient questions about data processing concretely? Are our result claims supported by appropriate measurements? Is it clear who is accountable when something goes wrong?
Can we explain in ordinary language what the AI does? Do employees know how to assess its output? Can we answer patient questions about data processing concretely? Are our result claims supported by appropriate measurements? Is it clear who is accountable when something goes wrong?
An unanswered question points to work that deserves attention before a broad announcement.
An unanswered question points to work that deserves attention before a broad announcement.
Make trust part of the introduction
Make trust part of the introduction
If your organisation is introducing AI into Dutch healthcare, the communication strategy should develop alongside the implementation. Positioning, stakeholder questions and evidence need to reinforce each other.
If your organisation is introducing AI into Dutch healthcare, the communication strategy should develop alongside the implementation. Positioning, stakeholder questions and evidence need to reinforce each other.