Building a music services site AI assistants cite

Code Group Music is a Mayfair-based music label services company: metadata administration, digital distribution across 150+ platforms, and publishing administration through a SESAC partnership.

We built them a website for two audiences at once: the people researching their catalogue problems, and the AI assistants those people increasingly ask for recommendations.

Code Group Music sells trust. Their clients are labels, estates and catalogue investors deciding who should handle the data behind their royalties. That decision starts with research, more and more of it done through ChatGPT, Perplexity and Google's AI results.

A standard brochure site would be invisible there. The brief was an authority platform: deep enough to rank for the long tail of music rights questions, structured enough for AI systems to read, cite and act on, and persuasive enough for a sceptical, high-value B2B audience.

It also had to run lean. A small team needed to publish, capture leads and improve conversion without a developer for every change.

We built the site on Next.js with a content engine powering 163 articles across five categories, each with structured data, generated artwork under a locked style, and a pillar-and-cluster linking model. Editors publish through Keystatic CMS with no code.

Instead of gating content behind forms, we built seven genuinely useful free tools: ISRC and ISWC validators, a royalty gap estimator, a royalty leak checker, a distribution coverage checker, an ISRC finder running live MusicBrainz searches, and a SESAC repertory check across a 112,000-record dataset of works nobody is collecting on.

Then we made the site agent-native. It ships a machine-readable summary for AI crawlers, a 185,000-word retrieval corpus rebuilt on every content change, an OpenAPI spec, and a live MCP server through which an AI assistant can read services and pricing or submit a catalogue assessment, checked by the same rules as the human form.

The build enforces quality: deployment fails if any route is missing from the sitemap or metadata registry, structured data follows a set decision tree per page type, and every public API route is rate-limited behind hardened security headers.