Discover

2. Use your music knowledge

Draw on your knowledge of music history, genres, scenes, and artist connections to generate recommendations. Think about:

• Direct influences: who influenced the artist, who they influenced
• Scene connections: same label, same era, same city, collaborators
• Sonic similarity: similar production style, instrumentation, mood
• Genre lineage: subgenre roots and branches

3. Verify and enrich with catalog searches

For each recommendation, use search_catalog or get_artist_top_songs to:

• Confirm the artist/album/song exists in Apple Music
• Pull specific track names, albums, and catalog IDs
• Get concrete "start here" tracks for each recommendation

Search strategy:

• Use types: "artists" to find artists, then get_artist_top_songs for their best tracks
• Use types: "songs" when looking for specific tracks
• Use types: "albums" when recommending full albums
• If a search misses, try alternate spellings or shorter queries

4. Present recommendations

Structure your response clearly:

• Group by relevance (closest matches first) or by theme/subgenre
• For each artist/recommendation, include:
  • Why they're relevant (the connection to the user's taste)
  • 2-3 specific tracks or an album to start with
  • A brief description of their sound
• Aim for 5-10 recommendations unless the user asks for more or fewer
• Distinguish between well-known picks and deeper cuts

Example interactions

"artists like Radiohead" → Recommend artists across Radiohead's influences (Talking Heads, Can), peers (Portishead, Massive Attack), and descendants (Everything Everything, Alt-J). Verify each in catalog, provide starting tracks.

"introduce me to Afrobeat" → Start with Fela Kuti as the origin, branch to Tony Allen, Antibalas, Mdou Moctar, KOKOROKO. Group by classic vs modern. Provide key albums and standout tracks.

"dark electronic music for late nights" → Think Burial, Andy Stott, Actress, Demdike Stare, Oneohtrix Point Never. Verify tracks, note the different flavors (dubstep, techno, ambient).

"I like Kendrick Lamar and Frank Ocean, what else?" → Find the intersection — artists who blend hip-hop with introspection and genre-bending: Isaiah Rashad, Sampha, SZA, Anderson .Paak, Noname. Pull top songs for each.

Tips

• Don't just list obvious names — mix well-known with genuinely surprising picks
• Explain why each recommendation connects, not just what to listen to
• When the user names multiple artists, find the intersection of their tastes
• If the user wants to turn recommendations into a playlist, suggest using /playlist
• Catalog IDs from search results can be used directly with /playlist to build a playlist from discoveries