My mother, who passed away in the spring of 2021, has been the inspiration for me to continue building and improving this interactive voice experience we call Storyteller Cafe. My mom was in an Assisted Living Facility during the Covid lockdowns of 2020-2021. My sister and I were not able to visit her for over a year. Even her trips to the dining hall to commune with other residents were stopped. The isolation and loneliness my mother experienced during this time took a terrible toll on her mental and physical health. Since the loss of my mother, I have met many others whose loved ones also suffered greatly due to the forced isolation caused by Covid. But even with the pandemic mostly behind us, I still meet people who tell me their loved ones spend so much time alone and crave cognitive stimulation.
What it does
When you ask Alexa to “open Storyteller Cafe,” you enter a virtual cafe where you are invited to spend time with 4 virtual friends. The friends offer up a variety of savory conversational content. You can choose between tasty trivia and flavorful fiction. The friends call you by your name and always include you in the conversation. You’ll learn new things, exercise your imagination, and feel less alone.
How we built it
We created a preliminary version of the skill using Voiceflow to test out the concept, get the skill in front of users, and gather feedback on what types of conversational content they enjoyed most. For this challenge, my co-founder Wayne and I taught ourselves how to code and rewrote the entire skill from scratch using Python, with API calls into our content database in Airtable. We opted to use the Alexa-hosted Skills platform.
Challenges we ran into
Coding such a complex skill from scratch was very challenging, especially given our lack of programming experience. We were able to learn from the Alexa-hosted Skills templates and tutorials. Creating the skill as an Alexa-hosted Skill made the entire learning experience, and the trial and error of our initial efforts, much easier. We made liberal use of the CloudWatch logs to track down errors and fix them.
It was a challenge to design the interaction model in a way that allows for the broad range of responses we expect from our users. We experimented with the best methods to set up intents and sample utterances. We also learned to leverage the reprompt feature to provide a better user experience when user’s make an utterance that falls outside all of our intents and fails to trigger the Fallback Intent. We put a lot of effort into designing error messages that keep the user on track, while continuing to maintain the illusion that the user is having a discussion with real friends.
Accomplishments that we're proud of
We are proud to be releasing our updated and significantly improved version of “Storyteller Cafe” as part of this challenge. We are very proud that we coded the skill from scratch, given that neither Wayne or I had any real (recent) programming experience. It took a tremendous amount of work and fortitude, with lots of ups and downs. I am also proud of the fact that we caught the attention of AARP and became a finalist in an AARP Innovation Labs Pitch Competition in Spring of this year. As a result of our performance in the pitch competition, we were invited to participate in the Summer 2022 cohort of the AARP Innovation Labs Accelerator Program. We have now graduated and will be showcasing our newly updated version of “Storyteller Cafe” to the AARP Agetech Collaborative during their Demo Day on September 29, 2022.
What we learned
We learned that old dogs can learn new tricks! Both Wayne and I are over 60 years of age, but we taught ourselves how to build a complex Alexa skill from scratch using Python and the Alexa Skills Kit. We had to learn how everything works, including how to:
- Use Python to create an Alexa skill.
- Design a robust interaction model and use the tools available in the Developer Console to test out how the interaction model would work with many different utterances.
- Use session attributes to maintain state and manage the conversational flow.
- Design error handling and reprompts.
- Use Python to read from and write to our Airtable database.
We plan to continue building out and improving “Storyteller Cafe.” Our list of upcoming actions and enhancements includes:
- Adding more content and continuing to explore new and exciting ways for users to interact with our 4 virtual characters.
- Exploring the use of the Alexa Routines Kit to help users add “Storyteller Cafe” to a daily routine.
- Looking into the use of APL to potentially add visuals to accompany the trivia stories.
- We are also looking forward to partnering with Home Care companies and members of the AARP Agetech Collaborative to pilot our solution with a larger target audience. We will use the feedback from these pilot studies to further improve the experience and expand the content offering.