ConnectAI

Facetime with a chatGPT-based AI therapist. You'll be surprised to experience how life-like it can be!

Comment

Inspiration

The inspiration to create a ChatGPT-based AI therapist comes from the need to provide mental health support and therapy to individuals who may not have access to traditional therapy due to various reasons such as cost, stigma, or lack of availability. The rise of mental health issues worldwide has led to an increase in demand for mental health services, and AI technology can help fill the gap by providing accessible and affordable support to those in need.

Additionally, the anonymity and privacy provided by an AI therapist can make it easier for people to open up about their thoughts and feelings without the fear of being judged or stigmatized. The AI therapist can also provide 24/7 support and can be accessed from anywhere, making it more convenient for individuals who may have busy schedules or live in remote areas.

What it does

ConnectAI is an AI-based platform that can be used as a virtual therapist to provide mental health support to people. It leverages the power of natural language processing (NLP) and machine learning (ML) to simulate human-like conversations with users and provide them with emotional support.

How we built it

We built the web application using Javascript, Python, HTML, and CSS. We have also made use of the following APIs to build the seamless pipeline: user voice to text -> Deepgram text to chatGPT & chatGPT to text -> OpenAI chatGPT text to audio -> Microsoft Azure

Challenges we ran into

  1. Most of the APIs were priced and hard to access
  2. Constant stream from speech-to-text to make the model understand when to stop and understand pauses in between was the most challenging part of all.
  3. Interoperability between JS and Python (All three of us are Python Devs)
  4. Managing async API responses
  5. Improving overall system architecture to make the pipeline smooth and latency-free

Accomplishments that we're proud of

  1. We've become way better engineers than we were while dealing with project management, collaboration, and problem solving.
  2. We've managed to become better prompt-engineers
  3. We've managed to integrate an entire speech-text-chatgpt-text-speech pipeline within 20 hours and we're extremely proud of how seamless the result turned out to be.
  4. We've also reached out to the founder of an API we were using that was expensive, and managed to get free credits for the hackathon under 2 hours.

What's next for ConnectAI:

Scale the system allowing concurrency and users to interact with the therapist with real life actions (calendar integrations and so on). Allow the user to select the preferred therapist based on character and look.

Share this project:

Updates