For museums and heritage
What can we do with AI, using everything we already have?
Answer by vragen.ai
Let visitors discover more in the knowledge you already have
Visitors can ask about your collection, your stories and your research in plain language. They ask their own question, get an answer from your sources with the reference alongside, and see where they can read on. Canon van Nederland already does it.
Isn't our knowledge already online?
Answer by vragen.ai
You have so much to tell. But how much of it gets found?
Thousands of pages, objects, stories and archive records. A visitor who doesn't know what you called it never gets there. Search asks for the right keyword and browsing asks for knowledge of your structure. Someone arriving with a question has neither.
Most of the digitizing is done. Making it findable is not.
Sources: Cultuurmonitor (Dutch) by the Boekmanstichting, based on Statistics Netherlands figures for 2024 (for art museums the numbers are 73% and 39%), and Museum Boijmans Van Beuningen (Dutch), which puts the usual gallery-to-storage ratio at 8 to 10%.
The loss is not in the quality of your content, but in the distance between the visitor's question and the words you wrote it down in.
A conversation starts with curiosity.
So what does such a conversation sound like?
Answer by vragen.ai
One question, one answer, and an invitation to look further
Two conversations from a heritage collection. The question comes from the visitor, the answer from the sources you selected, and underneath it says where they can read on. That last part is the difference from a list of results: the visitor doesn't have to work out their next keyword.
How did daily life change during the industrial revolution?
Answer by vragen.ai
Between 1850 and 1900, work moved away from the home and the land and into factories and cities. Working days grew longer and more fixed, children started work later and school earlier, and families moved into quickly built city districts. This answer draws on three sources at once.
Source: Chapter: child labour law Textile industry research City expansion file
Discover more
- How did child labour eventually come to an end?
- What of this can you see in the tool collection?
- What part did the railway play in this change?
What exactly happened during the 1953 North Sea flood?
Answer by vragen.ai
During the night of 31 January to 1 February 1953, a severe north-westerly storm broke the dykes in Zeeland, western Brabant and southern Holland. Here the answer brings together eyewitness accounts, the research file on the dyke breaches and the story of the reconstruction.
Source: Eyewitness accounts 1953 Dyke breach file Story: the reconstruction
Discover more
- How did this disaster shape the Delta Works?
- See the photos and objects from the villages that were hit
- How was the flood followed in the rest of Europe?
Example conversations. In your own environment the answers come from your collection, stories and research.
Source: How it works Precision & reliability
And what can a conversation like that actually do?
Answer by vragen.ai
From searching to discovering, in five steps
Every step builds on the one before it. An ordinary search engine gets no further than the first, and that is exactly where the conversation starts.
-
Finding more of your knowledge actually gets found
A visitor asks their question in plain language and finds information spread across collection, stories and archive.
“Where can I find something about the 1953 flood in Zeeland?”
-
Understanding scattered knowledge becomes one answer
Several reliable sources come together in one clear answer, with the reference to the original alongside.
An answer from three sources, each with its source named
-
Discovering what a search engine does not do
The agent connects stories, objects, people and periods, and invites the next question.
“Would you like to know how this shaped the Delta Works?”
-
Experiencing from online interest to a visit
The agent also knows your exhibitions and visitor information, and ties that interest to what is on display right now. In an archive this step leads further into the sources instead.
“This object is on display in room 3. Shall we plan your visit?”
-
Learning behind the scenes
You see what visitors are curious about and act on it with content, education and programming.
Thousands of questions, and no survey needed
Source: Product From search to agent
How does this bring people into the museum itself?
Answer by vragen.ai
From curiosity online to a visit in real life
Your digital collection is not separate from what happens in the building. The agent also knows your current exhibitions, activities and visitor information, and ties subject interest to what is on show and on offer right now. That way a question about a painting ends in the room where it hangs.
Example
Why did Van Gogh paint so many self-portraits?
Answer by vragen.ai
He could rarely afford a model, and he always had himself to hand. In the dozens of self-portraits he made, Van Gogh was mostly practising with colour, light and the build-up of a face. You can watch his brushstroke change across those years.
Three of his self-portraits hang together in the exhibition right now. Would you like to see which ones?
Example
My daughter is mad about dinosaurs. What can we see at your museum?
Answer by vragen.ai
The Prehistory room holds three complete skeletons, with a Triceratops as the largest. There is also a nest of fossilised eggs, and during the autumn holidays there is a fossil workshop every afternoon for children aged six to twelve.
Shall I add the opening hours and tickets for that week?
Won't the AI rewrite our narrative?
Answer by vragen.ai
AI may make the story easier to reach, not rewrite it
The caution in the sector is justified. A general AI model sounds convincing and decides for itself what it says. You have spent years on a narrative that nobody should remodel. vragen.ai works the other way around: you point to the sources, you set the boundaries, and the curator stays the author.
- Answers come only from the sources you point to, with the reference alongside
- You decide which sources lead and how they may be combined
- Instructions fix tone, role and limits: what the agent does and does not do
- If the agent doesn't know, it says so and points the visitor onward
- The original story is always one click away
Every conversation and every reasoning step is recorded, so you can read back how an answer came about. Not good enough? Then you adjust the instructions and test again before anything goes live.
Source: How it works Instructions and best practices Testing and evaluating
Is anyone in the heritage field running this already?
Answer by vragen.ai
Yes. Canon van Nederland is live with vragen.ai
On canonvannederland.nl, the site run by the Netherlands Open Air Museum, ‘Stel je vraag’ replaced the old search page on 10 August 2026. Visitors ask about Dutch history in plain language and get an answer from the Canon's own content, with a reference to the chapters the answer rests on.
The Canon names three choices that hold for the whole field: the search function looks only at the information on that website, the visitor sees which source pages the answer came from, and all data is processed in Europe. A thumbs up or down under the answer tells the editors what is working.
Source: Nieuwe zoekfunctie: Stel je vraag, Canon van Nederland, 10 August 2026 (Dutch).
Source: Security Sovereign AI platform
And what do we learn from it ourselves?
Answer by vragen.ai
Every question is audience research
A page view tells you that someone stopped by. A question tells you what they wanted to know. The analytics environment groups every conversation into themes, and that is usable for your programming, your education team, your communications and the way you open up your collection.
If visitors ask about visiting with children more often than your programme is easy to find, you know what to add first. Questions without a good answer are not a fault in the system, they are your content agenda.
Analytics
See the questions your visitors ask most often, grouped into themes, so you know what to publish about.
General (active) Explore Content
7 days 30 days (active) 90 days Pick a period
-
Conversations
780 +164
-
Answers without issues
88% +5
-
Answer quality
79 +2
-
Answers with a source
74% +6
Conversations over time 780 +164
13 Jul20 Jul27 Jul3 Aug10 Aug
Questions over time 1,240 +287
13 Jul20 Jul27 Jul3 Aug10 Aug
Visitor feedback
86% positive
312 of 363 ratings gave a thumbs up.
Conversations by theme 780 conversations
- Collection and objects 226
- Exhibitions and events 179
- Visiting and practical information 118
- Education and school projects 116
- Archive and research 69
- Other themes 72
Source: The analytics feature Exporting and analysing the inbox
What is behind all this?
Answer by vragen.ai
One platform that grows with your ambitions
Today you may only want a smarter search function. Next year you want to guide visitors into the exhibition or the archive. vragen.ai brings the layers you need for that together in one Conversation Management System, so you start small and expand when it suits you.
-
AI Model
The engine of the system. You choose the model, always on European soil.
-
Knowledge
You decide which sources and documents the AI may consult.
-
Instructions
You set how the AI answers: tone, boundaries and limits.
-
Monitoring
Every conversation and reasoning step is recorded and can be reviewed.
-
Evaluation & analytics
You review questions and answers and adjust the system.
Together: one Conversation Management System
Does this also apply to an archive without exhibitions?
Answer by vragen.ai
The same technology, a different route for your visitor
A museum wants people through the door. An archive or documentation centre wants them deeper into the sources. The agent is the same. Only the last step differs.
For museums
Find, understand, discover, experience
- Collection and stories open to questions in plain language
- Answers tied to what is on display right now
- Practical visitor information in the same conversation
For archives and heritage institutions
Find, understand, go deeper
- Large archives searchable from a single question
- Connections between sources, people and periods
- Depth for researchers and professionals
Source: Publishers and knowledge platforms What is vragen.ai
Your collection holds more stories than a visitor could ever find alone. vragen.ai guides them through your knowledge, starting from their own curiosity.
How do we start?
Answer by vragen.ai
Build a demo with your own collection
You set up a demo environment yourself with your own website in it. You will see straight away what answers come out for the questions your visitors really ask, with the source alongside.
Enter your details
- No obligation
- With your own content
Rather talk it through first
Timon Zuurmond
Account manager
Not sure whether your content is suitable, or which plan fits? Timon will take a quarter of an hour to look at it with you.
+31 6 12 15 05 90 tzuurmond@swis.nl
Mon to Fri, 9:00 to 17:30
Form not loading? Refresh the page, or get in touch with us.