TakeOff

An AI-assisted conversation learning tool

Comment

Inspiration

Everyone, to some degree, suffers from social anxiety. There may be no standard to gauge how adept one will be in adjusting to a certain situation, but with the advancement of AI, we can equip individuals with the tools they need to learn. TakeOff stands to bridge the gap between those willing to learn and the vast amount of resources available through AI. By gamifying the experience and encouraging the user through each step of the process, we seek a world where everyone can work towards building their own self confidence. “The Sky’s the Limit!”

What it does

TakeOff is a web-based platform that helps users practice speaking skills with the help of AI content generation. Upon submission of a prompt, a script as well as options for audio and video readings of the script will simulate a real-life social situation for the user to interact with. Our goal is to gently encourage users to break out of their comfort zone, hence the emphasis on finding and following one’s own pace.

How we built it

After designing a high-fidelity prototype in Figma, we started implementing our product using HTML, CSS and bootstrap, JavaScript, node.js, and the OpenAI API on Replit. Replit allowed us to see our application in real-time, seeing how it ran with the OpenAI API that we used for the chat completion functionality to create hypothetical scenarios in which users with general social anxiety might have difficulty navigating. (Please note that due to constraints, the actual product slightly differs from the prototype, which reflects our actual vision for the project)

Challenges we ran into

The biggest challenge that we ran into was implementing the AI in a way that would be more meaningful and specified than simply using ChatGPT. Since ChatGPT is still fairly new, we weren’t really sure how far we could push the limits of OpenAI API. Another challenge that we ran into was implementing the API. We had trouble finding the endpoint url that we needed and spent hours debugging to see which endpoint url that we needed.

Accomplishments that we're proud of

We walked into the hackathon without any previous knowledge or experience in implementing AI resources. We are proud of how far we came over the past 24 hours, especially how much we learned about AI and the respective skills this learning opportunity has presented us. Some specific examples include:

  • Adjusting to unique constraints of using AI
  • Applying constraints to AI for the sake of consistency throughout our website

What we learned

This project helped us learn a lot about AI and OpenAI, more specifically. We learned how versatile OpenAI was in that it could generate images, texts to chat with us, and allowing for large amounts of data to be easily processable. Even beyond OpenAI, we learned how to navigate technical problems as a team and how to overcome difficulties by becoming a support system for each other.

What's next for TakeOff

While the technology to solve many problems is available, it is not necessarily accessible. If given the proper resources, mentorship, and time, we hope to be able to launch TakeOff as a free OpenAI resource for everyone who wishes to improve their general speaking abilities. While the product we presented today was only able to implement images and audio, our original vision for TakeOff was to create a completely immersive experience for the user. We hope to be able to follow through with our original concept (fixing the UI and making it more user friendly) as we learn more about development with AI.

Built With

Share this project:

Updates