MeetAI | Devpost

Landing Page
System Architecture

Inspiration

In today's fast-paced environment, video calls often lead to important discussions and decisions. However, for hosts, generating new ideas to engage the audience can be challenging at times, and taking notes during a presentation can be inconvenient. Manual summarization often misses critical details.

What it does

MeetAI is an AI-powered video call platform that provides real-time feedback leveraging advanced technologies such as Fetch.AI's AI agents, Google's Gemini AI, SingleStore's vector databases, and Agora's real-time voice and video capabilities to transform how meetings are documented and actioned.

MeetAI generates a concise summary of the key points discussed, and a detailed list of notes, including key decisions made, action items with assigned responsibilities, important deadlines, and any unresolved issues or next steps all while the video call is going on. This feedback-driven cycle allows the host to make assisted decisions on how to continue to go on with the meeting.

How we built it

The front end of MeetAI is built using TypeScript and Tailwind.css to create a clean and minimalistic user interface. The backend is implemented using Python FastAPI to handle our API endpoints and SingleStoreDB for our vector database, allowing fast and accurate retrieval of information from large datasets. The Agora API is used to support our video streaming platform, which makes real-time communication easy. We also utilized the OpenAI API to generate text embeddings on our vector database and Fetch.AI agents simplify the process of generating notes and vector search.

Frontend: TypeScript, TailwindCSS Backend: SingleStoreDB, Python, FastAPI APIs: Gemini, Agora, OpenAI, Fetch.AI

Challenges we ran into

Accomplishments that we're proud of

We are happy to see that we have a beautiful landing page, providing a clean and smooth user experience. Combining so many new technologies in a short time was no easy feat, including the process of going through documentation and resources for vector databases.