👋 Hi, I'm Chiara!

Project background
Chiara is a chatbot that assists families during the referral process to the hospice. Chiara is the product of a large project in end-of-life care for children.

Commissioner
MAIS Family Office.

Team
Helix Centre: Ivor Williams (Project Lead), Chris McCormack (Project Manager), Jorge Alexander (AI Software Developer).

Responsibilities
User Experience, Visual and Interface Design, collaborating with engineers, managing user testing, research analysis & synthesis.

This post summaries the process of designing a chatbot called Chiara. It is an AI-powered chatbot that support families during the referral period to the children hospice.

I’d love to share some learnings. You will get insights about the process and decisions that can go into building a chatbot, and some references about the methodology and technology used.

OUR PROCESS

Let’s look at our four step process for building a chatbot!

  1. Envisioning the chatbot
  2. Designing the chatbot
  3. Building the chatbot
  4. Testing the chatbot

 
 

1. ENVISIONING THE BOT

We had already scoped the opportunity for an AI-enabled chatbot to help families caring for a child with life-limiting conditions at home. So the first thing to define is the exact goal of the bot in supporting a specific user segment of families who have never been to a Children hospice.

Chiara builds on an extensive research that our End-of-Life care team conducted last year. This phase of work helped us to shape the problem space and identify the needs of families caring for children with life-limiting & life-threatening diseases. To provide some context, some of the key insights from user research:

 

Goal

Defining the chatbot goal is making clear what people can expect from the virtual assistant. Our team identified 3 goals for Chiara. In order of priority:

  1. Providing recommendations for support
    The chatbot will make suggestions for support groups and resources tailored to family needs and context.
  2. Giving suggestions for Children hospices
    The bot will match families to the most appropriate hospice for them and hopefully shorten their journey to it.
  3. Connecting users with professionals at the Children Hospice
    Chiara acts as a connector between families and professionals, services and other families.

All of the above goals are highly personalised in user needs and context. Therefore they all would require two preceding steps:

  • Getting to know the family 
    The chatbot gathers information about the user's and family's preferences and introduces its role and capabilities.
  • Understanding disease and context
    The bot digs deeper into the health conditions and issues of the child in need.

 

2. DESIGNING THE BOT

Flow

Together with Jorge, our developer, we framed the chatbot functionalities by following a model called Prometheus. In which, we ask why (goals) before how (functionality descriptors). By using the Prometheus model, I was able to provide context and user goals & motivations to the developer.

Following the modelling, we created a tree-like task-flow on Miro. The task-flow shows four primary goals (in red), and a series of secondary goals (in green).

Chatbot personality

It's a concept commonly underrated on interfaces, but the bot's personality is a significant point to take into account as it defines how the bot will interact with the users.
We shaped Chiara's personality and tonality to fit users and their contexts: the bot should be transparent in its intention; it should sound confident and show an understanding of the user.

3. BUILDING THE BOT

Content

We build up a database so the chatbot could source its content. The database is a manually inputted spreadsheet in which we collect different resources:

  • recommendation resources appropriately tagged according to their content
  • a comprehensive list of all the Children's hospices in the UK
  • a living list of Children's diseases

 

Chatbot architecture

To iterate quickly, the developer leveraged the existing state of the art AI products and Web Services. We used Dialogflow, which comes with multiple benefits, such as its ability to integrate into various platforms, and the ability to analyse past conversations to improve future use.
Chiara required a visual interface to allow user testers to chat seamlessly and our team to experiment more quickly with different approaches. We chose for a web-based service called Kommunicate, which easily integrates with Dialogflow and provides some standard User Interface components – links, buttons, cards.

4. TESTING THE BOT

Testing the chatbot with families is fundamental, so we can understand how it performs under anticipated scenarios and how it works in real-world scenarios.
With families, we tested:

  • Chatbot personality: How well does Chiara express itself with language?
  • Chatbot Capabilities: How fitting are Chiara's suggestions to the user? To which extent are Chiara's resources appropriate to the user and their context?

 

Learnings from usability testing

We asked internal test users to have conversations with the chatbot to get initial feedback on the user experience and tech system. Testing is fun, and stuff always break! We kept note where the chat flow was breaking and kept on trying and improving it before doing usability testing with families.
We planned two testing cycles with families, both apart of two weeks so to give us the time to improve the first version of the chatbot and test it again. Unfortunately, usability testing in end-of-life care rarely goes as planned.

 
 
To accommodate families' needs and reach out to more significant numbers, we tend to run remote sessions. With user consent, each interview is carefully recorded and documented.

 
 
After testing session, we produce reports that include: the product goal, list of participants, the methodology used, and:

  • Product Feedback
    This section shows the participants' response to the prototype that we are building. The feedback is going to inform the whole team and lead the direction of the prototype. We often include visual materials (screenshots, video snippets and photos) from the testing and pulled quotes of the participants.

 

  • Key Observations
    We gather inputs and observation on and around the prototype. The information could come while testing the prototype or having an informal discussion, and it helps us to create a collection of ideas on how to extend the prototype.

RESULTS

Despite some minor language frictions, participants had no problem understanding Chiara's requests. On the opposite side, the chatbot requires a bit more training and fixing to understand the users fully.

Overall we had a great response from families and found out many other ways Chiara could be of help for them.

SOME CONSIDERATIONS

🤖 Designing a chatbot was a completely new challenge for me, which I enjoyed. I feel more aware of the level of details and the number of decisions that goes into building a chatbot.

🤝 Amongst the challenges of this project, the most unexpected was finding the 'right' process to collaborate with Jorge, our developer. Designing and building chatbots it is still a new field in which developers do the majority of the great work. As far as we have explored, there were not that many examples of collaborations between designers and developers we could follow. After some trials and error, we put a semi-solid process in place, but still, some steps are a bit undefined (e.g., who own decisions that are cross-discipline?).