Overview
BeatBot is an AI-agentic music composition platform that uses multiple AI agents working together to understand the user intention and create music tracks.
The Problem
Existing music generation models often involves training large models which can be computationally expensive and time-consuming. With the power of AI agents, we are exploring the potential of LLMs for music generation.
The Solution
Implemented an agentic workflow using LangGraph where multiple specialised AI agents collaborate to generate music tracks. The output is in ABC notation which is pure text, we integrated into a real-time renderer to allow the user to preview and edit the generated music, which is a more human-centered approach for human-computer collaboration.
Challenges & Learnings
There are a lot of challenges in this project, but the biggest one is: agentic AI music generation is a new field with limited research and resources available, we have to come up with creative solutions.
Please refer to the article The Evolution of Our Agent Architecture for more details.
I never deployed a web app before, so I had to learn how to deploy a web app too.
Please refer to the article From Localhost to Cloud Deployment for more details.