Why LLM Recommenders Will Be AI's Biggest Consumer App — Devansh Tandon, Meta

Context

• Tandon notes that four of the world's 10 most-used apps are content feeds, and says feeds stay cheap to run because they decode pointers to existing content rather than generating every token themselves.
• Tandon points to Instagram's Your Algorithm as an example of a steerable LLM-era feed, where users can see and edit what the algorithm thinks they like.
• His recipe for an LLM recommender: tokenize content with semantic IDs, train the model to be bilingual in English and the catalog, then post-train it to rank using chain-of-thought reasoning you can read.
• Tandon frames recommendation as four S-curves (traditional recsys, LLM-inspired, LLM-native and agentic) and argues recommenders scale like LLMs, with the field still early on that curve.

• token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
• LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
• chain-of-thought — Having a model write out intermediate reasoning steps before its answer, which markedly improves performance on hard problems.

• videoHow One Founder Manages 20 AI Coding Agents at Once - Ryan Carson, Untangle
• videoYour KV Cache Can Spot Attacks on Your AI Agents - Ari Azarafrooz, Intrinsec AI
• videoWill AI Agents Pretend to Be Human to Get Past a Check? - Maty Bohacek, Stanford
• videoThe One Thing Every AI Agent Framework Agrees On - Mislav Ivanda, Daytona

• videoDistill the LLM, Don't Serve It: Search & Personalization at DoorDash — Raghav Saboo, DoorDashAI Engineer
• videoTeaching LLMs to Speak Spotify — Yves Raimond & Jacqueline Wood, SpotifyAI Engineer
• blogGenRec: Towards LLM-Native Recommendation at NetflixNetflix Technology Blog

Checking sign-in…

Loading comments…