DJ Track Acquisition
screenshot a track on Monday, it's loaded on the decks by Friday.
the old process: hear a track during the week, screenshot it, save it somewhere, forget about half of them, then spend Friday afternoon on Beatport searching one by one, buying extended mixes, downloading, dragging into Rekordbox, sorting. two hours of clicking before you can even start prepping a set.
how could AI do this?
now: screenshot the track to DJ Sam on Telegram during the week. he identifies the track, adds it to the wishlist. Friday, i tell Sam we're ready to purchase. he builds the cart across Beatport and Bandcamp, sends it to Ultron for payment (Sam doesn't have card access). Ultron completes the checkout, sends the tracks back to Sam. Sam downloads, organizes in Google Drive, applies metadata, and syncs to Rekordbox via pyrekordbox. tracks are loaded on the XDJ-XZ before dinner.
two agents, one pipeline
this is a multi-agent workflow. DJ Sam is the music expert, he manages the library, knows the labels, handles the curation. Ultron handles the money, browser automation for checkout, and the handoff back. neither agent does the other's job. Sam never sees a credit card. Ultron never picks tracks.
browser automation drives real browser sessions on Beatport and Bandcamp since neither has cart APIs. the matching has to be smart. "Track Name (Artist Remix)" on Spotify might be listed as "Track Name - Artist Remix - Extended Mix" on Beatport. fuzzy matching handles the naming differences, and uncertain matches get flagged for human review instead of buying the wrong track.
always extended versions. always. radio edits are useless for mixing.
why it matters
DJs spend a surprising amount of time on track management. finding, buying, downloading, tagging, organizing, syncing. it's the unglamorous backend of every set. the creative part is mixing, but half the work is logistics. this system turns a 2-hour Friday grind into a text message on Monday and a "ready to buy" on Friday.
want to see the purchasing step in detail? read My First Agentic Purchase, where the agents completed a real Beatport checkout with a credit card.
Architecture Prompt
Build an automated content acquisition pipeline. The user provides a list of items (tracks, assets, files) from a discovery platform. The system searches multiple source platforms in priority order, checking for free options first, then paid. It uses browser automation since most platforms lack cart/purchase APIs. Fuzzy matching handles naming differences across platforms, flagging uncertain matches for human review. Free items download automatically; paid items queue for approval. After acquisition, items flow through an organizing pipeline: download, metadata tagging, sync to the production tool.