AdaptlyAI

Inspiration: In the current era of online learning, it's become evident that one-size-fits-all educational content often misses the mark. We believed that to make learning more effective, it should be tailored to an individual's personal experiences, preferred explanation styles, and language. We took inspiration from everyday interactions, where explaining complex topics using analogies from one's hobbies or experiences makes understanding a breeze.

What it does: AdaptlyAI listens to users as they share about their personal experiences, hobbies, preferred language for learning, and their favored explanation style. The application then processes this information and personalizes subsequent educational content to the user, making analogies and references based on their shared experiences. It ensures that every learning experience is unique, personalized, and most importantly, effective.

How we built it: We integrated Google's Speech-to-Text API to convert user voice inputs into textual data. This data is then passed onto the GPT-4 model, which processes the transcript to extract key information about the user's experiences, learning preferences, and language. The Flask backend manages the processing, while MongoDB stores the extracted data for future reference.

Challenges we ran into: Fine-tuning GPT-4 to extract concise and relevant information from the transcripts was a challenge, given the diverse nature of user inputs. Ensuring the audio files met Google Speech-to-Text's requirements meant additional audio processing steps, which initially were not anticipated. Managing and securely storing the user data while ensuring quick access for personalization. Accomplishments that we're proud of Successfully integrating multiple APIs (Google Speech-to-Text and OpenAI's GPT-4) into a cohesive system. Creating a unique onboarding process that doesn't rely on traditional forms, but rather an interactive voice-based interaction. Achieving a high degree of personalization in a prototype phase.

What we learned: The intricacies of voice processing and the importance of ensuring audio meets specific requirements for transcription. The power of language models like GPT-4 in extracting meaningful information from diverse user inputs. The value of user-centric design in creating impactful educational tools. What's next for AdaptlyAI Expansion of the content library to cover a broader range of topics. Implementation of a feedback loop to continuously improve the personalization based on user interactions. Development of a user-friendly front-end to make the platform accessible to a wider audience.

Built With

• google-cloud
• mongodb
• openai
• opencv
• python
• react
• tensorflow