Spotify's NEO recipe adds semantic IDs for catalog items to an open weightsA model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.Full definition → LLM such as Qwen, then grounds those new tokens with the backbone frozen; the team says continued pretrainingThe first, biggest phase of building a model: training it on enormous amounts of text so it learns language, facts, and reasoning in general.Full definition → instead wiped out the model's language ability.
After groundingTying a model's answers to checkable sources — retrieved documents, live data, tool results — instead of letting it answer from memory alone.Full definition →, Spotify instruction-tunes the model across many Spotify tasks at once, and Jacqueline Wood reports that this multitask tuning even helped cold-start audiobook recommendations.
Spotify reports that 98% of generated semantic IDs come out valid even without structured outputForcing a model's response to match a schema, so downstream code can parse it instead of guessing at prose.Full definition →, a useful data point for teams weighing constrained decoding against plain beam search.
Spotify's generative personalization spans a steerable DJ, prompted playlists, an editable taste profile and personal podcasts, and one in four US Premium subscribers now uses the system daily.
Terms in this piece · Glossary
grounding — Tying a model's answers to checkable sources — retrieved documents, live data, tool results — instead of letting it answer from memory alone.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
pretraining — The first, biggest phase of building a model: training it on enormous amounts of text so it learns language, facts, and reasoning in general.
structured output — Forcing a model's response to match a schema, so downstream code can parse it instead of guessing at prose.
Why it matters
Spotify engineers describe grounding new semantic-ID tokens onto a frozen open-weight LLM backbone to preserve language ability, a technique that let them cold-start audiobook recommendations and raise judge-human agreement by 91%.
videoDistill the LLM, Don't Serve It: Search & Personalization at DoorDash — Raghav Saboo, DoorDash
videoWhy LLM Recommenders Will Be AI's Biggest Consumer App — Devansh Tandon, Meta
videoWorld Models Need Causality, Not Pretty Pixels — Christopher Manning, Moonlake AI
videoRobotics Has Been Stuck for 70 Years — Deepak Pathak, Skild AI
articleEncoding Your Domain Expert: The Context Layer Behind Spotify's Data AssistantSpotify Engineering