“Stańczyk” chatbot for the National Museum in Warsaw

by Stańczyk team

Efficient project process, or how five strangers got together to design a prototype of the NMW chatbot in 24 hours.

Hackart

In late May 2018, the National Museum held a hackathon on the theme of artificial intelligence and new technologies. Apps, games, bots, overlays or plug-ins — all forms were welcome. The event was open to the public and participants were selected on the basis of the answers they provided in the registration sheet. Out of over 150 applicants, the organizers picked the final 40 participants — 5 previously formed project teams and 15 independent participants who were supposed to meet for the first time and team up during the hackathon.

Before the event kick-off, museum employees outlined 5 major project challenges — issues that they unanimously considered the biggest struggles of the museum in terms of broadly understood communication. Two of them (the ones we ended up choosing) were as follows:

1. Encouraging social media users to interact with the museum
2. Facilitating online ticket sales and access to practical information.

These challenges were followed up by three target audience personas who were the dedicated subjects of our projects’ user experience.

Project challenges

The most burning communication issues that the museum has to deal with on a daily basis were studied by analyzing three cases of user experiences, presented from their perspectives:

1. Marta the Mother — I have a small child who requires a lot of attention and involvement, I have to stay vigilant at all times and keep instructing “don’t touch it”, “stop that”, “leave it”…
   Besides, a child has many more needs than an adult — “Mommy, I need to wee!”, “Mommy, I’m bored, when is this over?”, “Mommy, what does this man need that sabre for?” etc. That’s why the best thing for me is events dedicated to children, but I struggle to keep track of them or even to remember about them.
2. Adam the Tourist — this is my first ever weekend trip to Warsaw. I don’t have much time, but I’d love to see all the main city attractions. I need quick and direct information about current events so that I don’t waste my time on pointless tourist attractions.
3. Tadeusz the Rummager — I’m interested in the history of the 19th and 20th centuries, I have an extensive knowledge in that field and I keep track with bated breath of all the new museum exhibits from those periods. I could comfortably become a curator of many an exhibition. At home, I have an incredible collection of Polish paintings, in which the museum should take an interest if they don’t want to miss the opportunity of purchasing such important artwork!

The team

We all signed up separately: Karolina, Dawid, Asia, Hania and Mateusz. We were one of the three teams whose members hadn’t met before. A quick introduction at the table, a few questions about our professions and we soon realized we were a part of a multidisciplinary team consisting of:

● a designer — Hania,

● a marketing and project manager — Karolina,

● a community manager — Asia,

● a programmer — Mateusz,

● a UX designer — Dawid,

Photo by Michał Murawski

Such diversity promised a truly creative brainstorm and guaranteed a very clear division of tasks. We could sense that if we came up with a good idea, it was very likely that the rest would flow smoothly.

Each of us came to the hackathon with some idea or direction which the project could take. For example, Dawid wanted to use beacons, Hania — elements of gamification, Mateusz — an artbot, Karolina — an augmented reality app. However, we all realized that we shouldn’t be starting with our ideas, but rather with the user’s needs. So we decided to go through the whole process from scratch.

Research

1. Identifying the problem

We started talking to the museum employees and familiarizing ourselves with the museum resources that were available. We learned about the issues that the staff had to deal with on a daily basis and analyzed the prepared personas.

2. Benchmarking

We took part in lectures aimed to inspire us with innovative ideas by presenting the most interesting solutions used by cultural institutions from all over the globe. Thanks to the help of experts in many different fields — from cultural studies professors, experienced new technologies designers and programmers through to business coaches — we were demonstrated all aspects of a start-up within just a few hours.

3. Empathization

We were slowly becoming overwhelmed by the amount of knowledge and ideas until we finally decided to do a tour of the current exhibitions ourselves. At that moment, all the budding ideas that had seemed quite brilliant before, were confronted with the reality of the museum. What came to the forefront were the problems that the visitors are faced with and the confusion that they experience as users. Thanks to her knowledge of Polish painting, Hania told us a few interesting facts that completely changed the way we perceived those artworks. The research into the museum exhibitions was concluded with a hypothesis: “one decisive detail can change the way you perceive art”, which informed a lot of our decision-making process as to the choice of the final project.

Generating ideas

Even though each of us had a few good ideas up our sleeves, we decided to focus for three hours on brainstorming new ones and then to assess all of them in the following categories:

a) practicality,

b) accessibility,

c) technological issues,

d) marketing potential,

e) the business reality.

We had plenty of different ideas: an educational game that guides you through the exhibitions, information tags about the works of art using a Tinder interface, an improved visualization of the digital collections, an app which allows you to talk to a painting… However, we rejected most of the ideas after we asked ourselves the crucial question: “What problem does it solve and for whom?”.

Even the answers to our favourite hypothesis of the “decisive detail” seemed like we were offering something of little value to the user but demanding a lot of effort on their part to become accessible (installation, onboarding, using it in context).

After long deliberations, we came up with the idea of an information chatbot, which finally met all the above-mentioned expectations. The idea of a chatbot was based on a specific problem that was customer service via messenger, as visitors tend to send plenty of questions to the museum before or after their visit. The museum staff kept stressing from the start that they didn’t have enough time to properly answer all the individual messages sent by the users. So we decided to develop a chatbot that would be able to solve these problems.

Analysis

It was a Saturday afternoon. All the other groups had already chosen their ideas and were working on implementing them for a while. We got down to work. We were simultaneously breaking down the frequently asked questions, getting access to the person responsible for communication in the NMW, we were checking how other already-existing chatbots work, thinking about the potential development and, most of all, we were wondering how to prevent our chatbot from being just an informative automaton without a soul.

We were trying to create a different chatbot character for all the initially defined persona. We were considering using figures related to the museum, such as Stańczyk, Bona, Pomarańczarka, Boznańska, Malczewski etc. We were on a roll, generating plenty of ideas and potential scenarios. We felt like we were on the right track. Yet, we decided to take a step back and critically analyse the ideas we’d had so far against the fundamental questions: “for whom? why? how?”. We repeated the same process we used when generating ideas, we were assessing them while bearing in mind that “less is more”.

The “Stańczyk” project

Developing many different characters to answer questions was time-consuming, difficult in terms of content accuracy and made the chatbot overcomplicated, which we’d badly wanted to avoid. So we chose just one character that we considered the best for that role — Stańczyk.

Stańczyk is one of the most famous artworks that the museum has in its possession, but also a very distinctive figure. He was considered a highly intelligent patriot with broad knowledge and an extraordinary sense of humor. That was the set of characteristics that we found to be ideal for a spokesperson of the museum. Moreover, his use of Old Polish language gave the chatbot a historic character, which not only suits such an institution perfectly, but also adds a funny twist.

At this stage, we already knew that the product we were working on would focus on storytelling, on constructing a certain narrative and character creation. We were also content that in the case of Stańczyk, technological barriers weren’t an issue (we used Chatfuel, a simple and free tool for programming chatbots). On that note, the first day of the hackathon came to an end — we left the museum at 10 pm.

The prototype

On Sunday, we started by creating our character — Stańczyk’s persona. Hania and Dawid were in charge of visuals, while Asia and Karolina dealt with the language side. At the same time, Mateusz was testing the possibilities of Chatfuel.

Then, we designed on paper a preliminary mock-up with a mapping of users’ paths, that is, the most frequently raised topics in the conversations between the NMW and messenger users. We chose to develop a prototype of:

a) an event query path,

b) an authentic case of a query as described in “Tadeusz the Rummager”,

c) witty ripostes to any attempts to catch the bot out.

We managed to create a clickable trial version of “Stańczyk”.

Our bots avatar looks like the Stańczyk from Jan Matejko’s painting but wearing call-centre-like headphones. Our Stańczyk always begins the conversation in a characteristic, Old Polish fashion: I give thee welcome, Sir! “I, the last of the Jagiellonian jesters, am a fountain of humour and knowledge. I shall help thee in thy need. He sometimes answers via gifs, such as Stańczyk facepalm.

Our priority was to build our chatbot MVP, which included:

1. Answering the most frequently asked questions
2. Suggesting upcoming events
3. Engaging users
4. Fighting against trolling in a constructive way

And so, after 24 hours of work, a prototype of the “Stańczyk” project was complete and ready to be carried out, rested and implemented. The project was very well received by the hackathon’s jury and considered as one that could truly be implemented into the museum. Thanks to that, we received the second prize in Hackart’s final, organized by the National Museum in Warsaw. It motivated us to continue working for free on developing the Stańczyk chatbot after the event ended. The execution phase is a whole other story, which we will keep for later.

Perspectives

Stańczyk is not only an information chatbot whose task is to instantly respond to users’ queries. It has an immense marketing and promotional potential for the museum. It can certainly create a positive image of the NMW, make the museum more accessible and engage its visitors to keep in contact with the institution through messaging.

What’s more, “Stańczyk” could one day become a virtual tour guide of the museum exhibitions inspired by our hypothesis of “one decisive fact”. It can also act as a private assistant, sending out personalized information about the museum events or it can even turn into a Stańczyk podcast, regularly informing the public about the museums' resources. Finally, Stańczyk could also become an Ambassador of the National Museum of Warsaw.

Implementation

We believe that Stańczyk will prove to be a valuable product. We are truly motivated by how simple it is to validate the idea. Implementation of the Stańczyk MVP costs nothing and thanks to our efforts and the kindness of the Museum and Wandlee — company which partners with Facebook in creating chatbots, we can put it to the test in real-life conditions. By using the method of iteration, we will be able to tell whether the users interacting with our chatbot are satisfied. If the results are positive, it will be a clear sign for us, supported by real data, that developing the project further makes sense.

If it works out, our future goal would be to develop the chatbot’s personality through cooperation with historians, copywriters and interaction designers. Our final goal is to reach a point where Stańczyk’s narrative is so characteristic that it becomes an effective means of establishing engaging communication between the museum and the users.

Soon, we will write a bit more about the process of developing our gallant sir. ;)