MeetingMate

Inspiration

We are a team of five undergraduate computer science students who have spent many hours in online meetings for classes and internships. We know firsthand how difficult it is to be fully present in meetings, while maintaining productivity and communicating effectively. With this project, we wanted to create a minimalist, automated alternative to other note-taking apps with a suite of features to maximize in-meeting productivity.

What it does

The project is an innovative productivity tool designed to facilitate effective collaboration and communication during meetings. The app helps users to be fully present in meetings while maintaining productivity and effective communication. It incorporates automated language processing to extract important details from conversations and generate comprehensive summaries based on predefined templates, eliminating the need for manual note-taking.

For example, during a daily stand-up meeting, users can provide a specific meeting template that the app uses to accurately match the conversation with the corresponding topics. The same can be applied to various other meetings such as sprint planning, scrum meetings, and more, and the templates can be customized according to individual preferences.

Additionally, the app uses live transcripts to capture the entirety of the meeting, which can be reviewed later to generate actionable insights. This allows users to remain fully engaged in the conversation and derive valuable insights from the meeting discussions without the need to take detailed notes.

How we built it

We divided the work amongst ourselves based on our individual skill sets and interests. One team member focused on the front-end development, while another worked on the server-side, and two more worked on the language processing and Zoom SDK integration, respectively. The fifth team member assisted wherever necessary and provided valuable insights throughout the development process.

The front-end development involved creating a clean and minimalist user interface (UI) using the Svelte, Smelte, React & JavaScript library. We designed the UI to be straightforward, with only the essential features and information displayed to the user, to allow them to focus on the meeting at hand. The UI was also designed to be responsive to different screen sizes to provide an optimal viewing experience on any device.

For the server-side development, we used the MERN stack (MongoDB, Express, React (Svelte), Node.js) and deployed front end on Netlify and backend on Heroku. We developed a RESTful API to allow the front-end and other components to communicate with the server. The API was designed to handle HTTP requests and responses to ensure seamless communication between the server and client. Additionally, we used the ZoomSDK to host the front end of the app to the zoom client.

The language processing component was developed using the GPT-3 model provided by OpenAI. This machine learning model allowed us to analyze the meeting transcripts, identify the key topics discussed during the meeting, and generate a summary of the key points discussed.

Finally, we were hoping to use the Zoom meeting SDK integration component was developed using the Zoom Meeting SDK. We intended to use it to connect our app to a live Zoom meeting and retrieve real-time data, such as meeting participants, chat messages, and transcripts. However, because of technical challenges we were only able to host the front end on zoom but implementing zoom meeting sdk was unsuccessful.

Challenges we ran into

Developing an app can be a challenging task, especially when trying to integrate a third-party SDK, such as the Zoom Meeting SDK.

Despite the challenges we faced while working with the Zoom Meeting SDK, we tried our best to deliver the best product possible. We switched to React from Svelte, which proved to be a better fit for our needs, and we continued to push through the obstacles.

Accomplishments that we're proud of

We are extremely proud of what we were able to achieve during the development of our productivity tool. Despite the challenges we faced while integrating the Zoom Meeting SDK, we were able to get our individual components to work without having much development background.

Our backend and frontend were able to communicate seamlessly, and the GPT-3 language processing model was not only able to work but also provided accurate summaries of conversations. We were also able to deploy both the front and backends on Netlify and Heroku successfully, which was a significant accomplishment.

While the unresolved issues with the Zoom Meeting SDK posed some difficulty, we were able to learn valuable skills in debugging and problem-solving, which will undoubtedly be useful in future projects. Moreover, we learned how to work effectively as a team, contributing our individual strengths and working collaboratively to achieve a shared goal.

What we learned

During the development of our productivity tool, we gained a wealth of technical knowledge and interpersonal skills. Specifically, we learned more about the MERN stack and had the opportunity to use various cutting-edge technologies such as GPT-3 for language processing and the Zoom Meeting SDK for in-app video communication.

In addition to technical skills, we also developed important interpersonal skills such as effective communication, time management, and problem-solving. We learned how to navigate and resolve conflicts and how to work collaboratively as a team.

What's next for MeetingMate

In the future, we aim to enhance our app by successfully integrating it with multiple video conferencing platforms, including Teams, Google Meet, and others. We plan to achieve this by adopting a modular design approach, which will enable our app to easily integrate with multiple platforms.

Furthermore, we intend to leverage machine learning to create a model that is specifically tailored for software and technological fields. This model will help us to understand the nuances of note-taking and minutes generation in these fields, leading to improved accuracy and efficiency in creating agendas and minutes.

We also plan to build upon the current GPT-3 model and integrate it with our new machine learning model, leading to even more accurate and efficient results. We believe that this will allow us to provide a valuable service to professionals across multiple industries.

In conclusion, we are committed to continuing to improve and expand our app to provide maximum benefit to our users.