FAQ.

What is this?

Hi, I'm Andy. I rebuilt Discogs in March because I was bored and wanted the mobile site to work better: videos wired in, a fast API, and a trial run as an MCP server. The full catalog cost about $2k in database bills, so I parked it.Follow me here:

What is Dig now?

I slimmed this idea down to house and techno, with an LLM layer on top: query the Discogs catalog directly, chat about your collection and the artists you like, get recommendations with YouTube links. It works pretty well.Take it as an open experiment in what Discogs could be. They own an impressive dataset and can commoditise it in fun ways.Dig is demo only. Deliberately light. No logins. Runs on my API credits (Kimi 3). Request access here.The long version of the build is at how we built it.

Where does the data come from?

The monthly Discogs data dumps (CC0), cover art from the Cover Art Archive, crosswalks from MusicBrainz (CC0). The editorial layer is mine: the scenes, the core runs, the related-label edges.The search database needs a bit of tweaking (some tracks and artists are missing), but as a beta demo I'm happy with it.

Is this Discogs?

No. Independent, built on their openly licensed data. Pressing-level detail (versions, matrix numbers, marketplace listings) lives on Discogs, and every record here links straight to it.We stick to the Discogs ToS, so as yet we can't use their cover art or integrate your collection and wantlist. Boo.

Why is a record missing?

Either it's outside the scope (wrong genre or era, the catalog is deliberately narrow) or it landed since the last monthly dump. If it's in scope and still missing, tell me.

We also built a drum-pattern generator from some of these scenes — try it.