• 

  Couch Hero Website

• 

  Couch Hero on Messenger

• 

  Structure of current solution

Welcome to Couch Hero!

An AI helper to easily find all the answers to your corona-virus questions, match elderly people with volunteers in their community and connect individuals to existing solutions.

Inspiration / Problem

The United States is going to be hit significantly with the coronavirus. Around 40 % of the elderly population live alone in their household. 90 % of calls to 911 are currently related to COVID-19. There is a need for a solution offering domestic help, combating loneliness and raising medical awareness. Especially among those members of the community who are less comfortable with using technology.

What it does / Solution

No need to reinvent the wheel! There are a lot of different great solutions out there already, it is just hard to find them. Couch Hero is a chatbot that can answer the users' questions about the coronavirus. Couch Hero can also direct the user to resources most useful to the users' needs. It does that through a normal conversation and is able to be easily integrated into most social media platforms.

How we built it

The main functionally of couch hero is built using Google's Dialogflow, an AI for building conversations. Here the main text and speech recognition is done. We've made a Heroku server to handle the logic of fulfilling the users' enquiries once the answers are not static anymore.
The processing goes:

User input -> Dialogflow -> Heroku server -> response to the user

(See attached images for solution architecture)

Challenges we ran into

The first challenge we ran into was that none of us had worked with Dialogflow before, so we had to learn fast. From there the next big challenge was making Dialogflow and our server co-operate, as without the server we would only have static predefined responses.

Accomplishments that we're proud of

The making the server and Dialogflow co-operate, giving us the ability to implement almost any (programmable) solution. Further, making it possible to call the chatbot by phone which is especially interesting for the elderly.

What we learned

As it was the first time for all of us to use Dialogflow, we all got the chance to develop knowledge in how to interact with users through an AI chatbot. Besides, it strengthened our skills in learning quickly. Additionally, as most of us didn't know each other and having limited hackathon experience, we learned to organize and communicate efficiently. Moreover, we learned how important it is to define the target group early in the process so that the features can be drawn from there.

What's next for Cytotoxic Tees

We have a few next steps to improve the product and to really make it deliverable:

• Add more questions and answers to the bot
• Train the chatbot so it recognizes the questions better
• Include existing solutions to refer the user to
• Improve the security
• Create a go-to-market / marketing strategy
• Promote the product

Integrations

* For free calls to the US, you can use the app TextNow.

Features

End-users can

• Get answers to FAQ regarding corona-virus
• Volunteer themselves to shop groceries for others
• Request that someone shop groceries for them
• Get updated on current epidemiological statistics
• Be referred to chatting and talk services if feeling lonely
• Get referred to official resources and questionnaires if feeling sick

Built With

• dialogflow
• facebook-messenger
• heroku
• python
• ruby-on-rails