The Summarizer project is an intelligent solution developed by Globus IT that automatically transforms meeting recordings into text, highlights key points, and generates concise summaries (meeting minutes). It's a tool designed to save teams time, ease access to information, and simplify post-meeting documentation.

In nearly every project, teams rely on video calls and online meetings. But rewatching hour-long recordings just to extract the main points is inefficient. To optimize this process, we needed a solution that would:

• Automatically transcribe video content
• Generate structured meeting summaries
• Highlight key points (meeting minutes)
• Easily integrate into existing IT systems
• Ensure data confidentiality

We developed a web application with a user-friendly interface and robust architecture:

• Frontend: Built with Next.js for a responsive UI and fast performance
• Backend: Based on Node.js (Express.js) and FastAPI, enabling flexible request handling and easy scalability
• Audio Transcription: We used OpenAI’s whisper-1 model, one of the most accurate transcription models on the market
• Preprocessing: Audio files are preprocessed using pydub
• AI Workflow: LangGraph is used to build AI agents that sequentially process the text, extract key fragments, and generate summaries and meeting minutes
• Content & Admin: Managed via PayloadCMS, providing seamless content handling for the client’s internal team

1. Video processing time reduced from several hours to 10–15 minutes
2. Up to 80% time savings for employees preparing meeting reports
3. Full automation of routine documentation tasks
4. Easier access to meeting content — key points and summaries are available right after processing
5. The solution is easily scalable and integrable into any corporate environment