While apps like RefAid and other websites successfully provide Refugees all the information to address their most basic needs (food, shelter, health, legal, etc.) we realized Refugees also suffer distress and mental health conditions which are difficult to address without the support of friends and family or of a mental wellness expert.

Refugees don’t have a space to talk about their problems, especially while they are on the move. For example, they might be depressed but don’t feel they can speak to those around them because those people are also facing tough times.

To address this problem, we created Sayfe Space. Sayfe Space gives Refugees a platform to naturally describe their problems and get support by anonymously chatting with volunteers that can empathize with their situation and are keen to offer support.

What it does

Sayfe Space offers a conversational UX to the Refugee, initially allowing them to chat with a chat-bot that using NLP (through IBM Watson) understands the condition of the Refugee.

While offering basic support through static information available in its database (e.g. addresses of the closest clinics), the chat-bot asks questions to form an anonymous profile of the Refugee to match him/her with the best suited volunteer to support their needs.

After the volunteer is identified, the Refugee is asked to confirm if he/she wants to chat directly with the Volunteer and eventually a direct chat is established.

How I built it

We used IBM Watson for NLP and the chat-bot and PubNub for the messaging protocol.

Challenges I ran into

Integration between IBM Watson and the UI as well as PubNub

Accomplishments that I'm proud of

We managed to have a functioning chat-bot that is able to invite another user (i.e. Volunteer) into the conversation based on it's understanding of the main user (i.e. Refugee) problem.

What I learned

Training an AI is more complicated than we thought and still requires a lot of integration with a nice UX/UI.

What's next for Sayfe Space

Improve the UX/UI, discuss the solution in more details with TechFugees to refine it and consider expansion beyond Refugees use case.

Share this project: