ThinkWellAI

Inspiration

The inspiration behind ThinkWellAI came from the desire to provide accessible and personalized mental health support to individuals worldwide, especially in areas where access to traditional therapy may be limited. ThinkWellAI is your personal therapist and a private place to share your thoughts without judgements at the least

What it does

ThinkWellAI is your personal guide on the path to wellness. By remembering the conversations and drawing from the proven techniques of Cognitive Behavioral Therapy (CBT), it creates a tailored plan to help you manage stress, anxiety, and more. Think of it as a gentle nudge in the right direction, empowering you to take small steps towards a happier, healthier you.

How we built it

The key innovation we had was that since transcripts of therapy counseling sessions were so protected and hard to get, we focused on finding well-defined guides to conducting Cognitive Behavioral Therapy and converting that into a knowledge base for the LLM along with a prompt based on these techniques. We then also use logic chains by making the LLM generate a plan and case summary at the end of each session which it then uses to start the next session.

Challenges we ran into

We were all backend people so coming with a frontend and connecting it to our backend turned into a nightmare. In the end given that we spent so much time training the LLM, we decided to simply use a streamlit platform. However we do have an implemented frontend backend system but were not able to complete everything on that system in the time limit

Accomplishments that we're proud of

We took a creative route to trying to work in the field of LLM therapists and we can actually see some results! Additionally we were able to keep the LLM on track through multiple sessions using Plans and case summaries which ended up being very effective.

What we learned

We have learned how to finetune LLMs, use RAG, prompt engineer. We have learned how to use streamlit. We have learned flask APIs and React JS. And finally we have learned what a hackathon is like as this was out first time.