GEM - A virtual assistant
During the start of my second year, we began a half year long project requested by the municipality of Rotterdam. They had been developing a virtual assistant called “GEM” for their website, to help residents with tasks such as finding information about passports, residency, travels, and much more.
For our project, they had one important question: “How do we make this product more accessible and inclusive?”
A new take on interaction
During the project we quickly came to realize a few key things. GEM did not have a lot of ways for the user to interact with it. Features such as configuring it’s language were not yet present, and the only way to converse with the chat bot was through text. This posed some issues for accessibility, particularly when it came to the focus group we had selected: people with low literacy.
We started the project by adding one simple feature: allow GEM to read out it’s own text. It’s a start, but it doesn’t solve the issue yet.
After various prototyping sessions, brainstorms and a design sprint, we came to another idea. The way GEM uses language should be flexible – there is no need for a chat bot to adhere to rigid and often times difficult language for our focus group. We started thinking of new ways to allow users to interact with this assistant in ways that worked for them. One of the core features was to create an option for GEM to “simplify” it’s language. This feature helped people who weren’t as experienced with the Dutch language to understand GEM more easily.
Features
Aside from changing the wording GEM uses and the way it forms sentences, we decided GEM should also be more than a chat bot. It should be a tool that changes the way someone can interact with the website – and not just a chat bot relegated to a corner of the screen.
We started working on this by broadening the ways GEM is applicable within the site – adding functionality to allow users to drag the assistant across the site and use it to search for articles, translate pages for them or read out specific segments. Another core feature was that we made it easy for users to share specific articles through WhatsApp for example – as it was common within our focus group for people to ask family or friends for assistance when they ran into issues.
Prototype
In the end we managed to create a high-fidelity prototype that turned GEM from a simple chat bot that was relegated to the corner of the screen, into a fully fledged virtual assistant that could help users navigate through complex articles and assist them in whichever way they needed.
Not every feature we originally had planned for made it into our final version, despite how impressive they were or cool to interact with. Because ultimately, we were making something simple, something that would help our focus group, and not reinvent the wheel entirely.
Our prototype for the virtual assistant “GEM” can be found through the button below!
