Skip to content
All systems
02Experimental

Kavosuri

AI-powered social music platform

Overview

An experimental platform combining generative AI and social music - lyrics to songs, remixing, and a feed built around AI-made tracks.

The Problem

Generative music tools are powerful but isolated - a person generates a track and it goes nowhere. There's no social layer built around AI-made music the way there is for AI-made images or text.

The System

Kavosuri takes a lyric or idea, generates a full track across multiple styles, and puts it into a social feed where tracks can be remixed, shared and streamed. The generation pipeline and the social product are built as one system, not bolted together.

The AI / Automation Layer

Generation is orchestrated through Vertex AI and audio/lyric-generation models (Suno, Lyria) behind an automated content pipeline - a request comes in, gets routed through generation, post-processing and moderation, and lands in the feed without manual intervention.

The Outcome

A bold, experimental product that treats AI generation as infrastructure for a social experience, not the whole product.

Future Direction

Deeper remix chains - tracking lineage between an original generation and everything remixed from it.

Technology

  • Suno
  • Lyria
  • Vertex AI
  • Python
  • API Orchestration
  • Next.js