About MindfulMate

Hi! I am Matt and I created MindfulMate, a mental wellbeing planner powered by GPT-3.5. With this, anyone will be able to generate a week's plan of personalised activities that can improve mental wellness based on their feelings and associated symptoms. Currently, MindfulMate is a Telegram bot that is available to anyone that scans the QR code above.

This project started during my military service when my friends and I were inspired by a unit effort that gave servicemen steps on how to overcome various mental wellness challenges that they might face. However, this was not a long lived effort as everyone was different, and what applied to one person might not neccesarily apply to another.

This posed a question: could something be made that creates personalised plans for people to work on their mental wellness? It was also around this time where I was playing around with Large Language Models and GPT. My curiosity to find out how far an LLM could help mental wellbeing led me to giving this idea a shot - and thus, MindfulMate was born.

Currently, I am the main maintainer of MindfulMate, coming up with features and refining the backend. However, being in National Service takes up a lot of time and that means I can't do everything. Jasper has stepped in to help me with bringing this project to life and I am extremely grateful for everything he has done, and is doing!



We take privacy seriously, and at the moment no user input is permanently logged or stored.

We genuinely hope that MindfulMate is able to give you relevant plans that can alleviate stress and tension to give you greater clarity throughout your week, and feedback is greatly appreciated.

If you would like to have a chat about MindfulMate, feel free to connect with me on LinkedIn here and I would be more than happy to have a conversation! You can also reach Jasper at his LinkedIn here.

Future Plans

We hope to collect your feedback and learn from the experiences of users! First and foremost, we hope to upgrade to using GPT-4 for our model to create more accurate and relevant plans for users. On top of that, we want to log any potential bad or unsatisfactory responses that we can use to improve MindfulMate.

We are also looking into creating a way to remember outputs from users so that additional follow up plans can be generated, building up on previous and creating a cohesive experience.