Case Study

Spheretrax tagged its entire soundtrack catalogue, automatically, in 8 weeks.

Using Cloud202's Qubitz AI platform, Spheretrax deployed a production-ready Automated Tagging Engine that reached >90% metadata accuracy, cut manual tagging effort by 70%, and went live in just eight weeks.

Customer Profile

Company: Spheretrax Limited

Industry: Media & Entertainment - Music Technology

Location: Hitchin, Hertfordshire, UK

Company No.: 14087534

Web: www.spheretrax.com

Business: Next-generation media intelligence platform for metadata management, soundtrack ingestion, and royalty operations.

“Cloud202's Qubitz AI platform transformed how we manage our soundtrack metadata. Their Bedrock-powered tagging engine delivered accuracy and speed we simply could not achieve manually.”

Sefi Carmel

Chief Executive Officer - Spheretrax

The Challenge: A catalogue growing faster than humans could tag it.

As Spheretrax's soundtrack catalogue expanded across diverse genres, licensing models, and commercial use cases, several operational bottlenecks emerged. Manual metadata tagging was time-consuming, inconsistent, and lacked contextual depth. Every asset needed human review to assign mood, genre, tempo, and licensing tags, introducing errors and delaying time-to-market for campaigns.

Ingestion workflows for digital marketing and back-office finance were fragmented. Submitting tracks, validating metadata, and routing assets to downstream systems meant disconnected tools and manual handoffs, slowing campaign launches and creating reconciliation problems in royalty workflows.

Partner and vendor data exchange compounded the problem. Without standardised APIs or automated validation, onboarding new partners was slow and error-prone, creating compliance risk and limiting commercial scale. Spheretrax could not grow its catalogue or its partner ecosystem at the pace the market demanded, without fundamentally rethinking how metadata was produced, managed, and exchanged.

Why Spheretrax Chose Qubitz: A production-ready agentic platform, not infrastructure to build.

1. Proven AWS-Native AI Platform - Built on Amazon Bedrock and Bedrock AgentCore with pre-built RAG, Knowledge Base, and AI orchestration. Faster implementation, lower risk, no AI infrastructure to stand up from scratch.

2. Meaningful AI With Human-in-the-Loop - Intelligent agents paired with human oversight. Every output is accurate, trustworthy, and business-relevant, with governance and compliance preserved throughout.

3. Enterprise-Ready and Scalable - Secure, resilient, multi-account AWS architecture designed for reliable production operations and seamless commercial growth.

Cloud202 delivered across three workstreams over four 2-week Agile sprints, with sprint demos and stakeholder sign-off at every stage.

What We Built: Three workstreams, one coherent platform.

1. Automated Tagging Engine

An LLM-powered engine on Amazon Bedrock's Nova model. A RAG architecture with a PGVector-backed Bedrock Knowledge Base retrieves licensing rules, genre definitions, and historical usage before generating contextually accurate tags, delivering >90% accuracy without manual intervention.

Bedrock Nova - RAG - PGVector - Knowledge Base

2. Data Exchange & Platform Management

An API-first data exchange built on Amazon API Gateway and AWS Lambda for secure, standardised metadata sharing. Automated ingestion pipelines, validation workflows, and partner onboarding replaced manual processes, cutting onboarding time and ensuring quality from the point of submission.

API Gateway - AWS Lambda - API-first

3. Soundtrack Ingestion Patterns

Real-time and batch ingestion on Amazon DynamoDB and AWS Amplify, integrated with CRM, DAM, and royalty systems. A Next.js frontend on Amplify gave the team a responsive, secure interface to manage ingestion health and campaign readiness in real time.

DynamoDB - AWS Amplify - Next.js - CRM - DAM

Return on Investment: Outcomes, not activity.