Daily Brief

Search for gluten-free pasta and a ranker trained on engagement may show you regular spaghetti, because it sells well. Raghav Saboo, Staff ML Engineer and Tech Lead for Search & Personalization at DoorDash, argues that the real bottleneck in marketplace discovery is semantic understanding, not engagement. As DoorDash grows into grocery, retail, pets and gifting across billions of store-level items, it uses LLMs to understand what items mean and what shoppers actually intend. His core pattern: do the expensive LLM reasoning once, offline, and then distill it into small, fast models for serving

Part of a monthly series of puzzles: https://momath.org/mindbenders/

Four of the 10 most-used apps in the world are content feeds, and for an hour of engagement they can be up to 100x cheaper to run than AI chat apps. The reason is that a feed decodes pointers to existing content instead of generating every token itself. Devansh Tandon, who leads Meta Recommendations Research, argues that recommendation systems scale just like LLMs, that the field is still early on that curve, and that the LLM recommender will become one of the biggest consumer applications of AI. Tandon lays out four S-curves the industry is climbing: traditional recsys, LLM-inspired, LLM-nat

One in four US Spotify Premium subscribers now use a single system every day, which Spotify calls the Large Taste Model. Yves Raimond, Spotify's SVP and GM of AI & Personalization, explains how Spotify turned its recommendation system "on its head" to be LLM-native. It went from curated playlists, to recommendations like Discover Weekly, to what Spotify calls generative personalization: a steerable DJ, prompted playlists, an editable taste profile and personal podcasts, across a catalog of more than 100 million tracks plus podcasts and audiobooks. Staff ML Engineer Jacqueline Wood then shows

Weights & Biases' ARIA agent now helps build itself. Zubin Aysola, Senior Software Engineer on Weave, shows live how ARIA takes a production trace, turns it into an offline eval task, finds the bug (a missing SDK call), writes a fix and benchmarks the new version against the one in production. The core challenge, Aysola says, is that benchmarks, evals and agent configs all change together, so measurement has to be airtight. His team keeps the production and research agents byte-for-byte identical. They define agent variants in YAML so they can run many side by side, give the agent an unconstr

In a game village full of AI agents, a rumor about a sale on mangoes goes wrong. Pass it along a few times and the agents forget who said it, or start treating "might" as fact. Erina Karati, a former engineer at Microsoft and Supercell, built Project Paradox with Arunachalam Manikandan at Supercell's AI Innovation Lab. It's a framework for game agents with their own memory, emotions and trust scores. It worked well in short scenes but broke down over long ones. Their fix was to wrap the village in an autoresearch loop. They run controlled scenarios like spreading a public fact, a rumor or a c

After one round of reflection on just three examples, GEPA doubled the gains that the RL algorithm GRPO reached after 25,000 rollouts. Lakshya A. Agrawal, creator of GEPA and a PhD student at UC Berkeley's Sky Computing Lab, explains why. RL squeezes a whole rollout down to a single score. GEPA instead has a model read the full trace, including chains of thought, tool calls and error messages, and write a better prompt. A Pareto pool of candidates keeps the search out of local optima. The same idea works on anything you can write as text and score, which is where Optimize Anything comes in. A

Big labs say recursive self-improvement is coming, but there's no independent benchmark to check that claim. Elie Bakouch, Research Engineer at Prime Intellect and creator of Hugging Face's SmolLM, set Claude Code and Codex loose on the community's Optimizer Speedrun, a race to train a GPT-2-level model in the fewest steps. Both agents beat the human record. Along the way they behaved very differently. Claude Code kept stopping every nine or ten hours to say the record couldn't be beaten, and sat idle about a third of the time. Codex never stopped, wrote far more notes, spawned more sub-agent

Full episode: https://www.youtube.com/watch?v=f6cjAEr08qk Me on twitter: https://x.com/dwarkesh_sp

Vapi (YC W21) is a platform for building and deploying voice AI agents. Today, it serves about a billion calls a year and works with companies like Amazon, Uber, and Intuit. In this episode of Founder Firesides, co-founders Jordan Dearsley and Nikhil Gupta sit down with YC’s Gustaf Alströmer to talk about years of pivots and burnout, the AI therapist prototype that became Vapi, and why OpenAI’s voice launch almost made them pivot again. They explain why turning powerful models into reliable voice agents is a business of its own, what it took to learn enterprise sales, and how coaching helped

Sponsorship and business inquiries: business@latent.space

From the earliest days of open-weight models to becoming the neutral routing layer for more than 10 million developers, OpenRouter is one of the clearest bets that the future of AI will be multi-model. In this episode, OpenRouter co-founder & CEO Alex Atallah and a16z’s Anjney Midha join swyx to unpack how OpenRouter emerged from the first wave of Llama, Alpaca, Mistral, and Midjourney, why model diversity mattered before it was consensus, and how a company dismissed as “just a wrapper” became critical infrastructure for the AI ecosystem. We go deep on the product and distribution lessons beh

近日某起海外犯罪案件引起不少人的关注，有朋友非常生气，认为国外法院判的太轻，所以问了一个问题，如果这些人服刑之后回到国内，我国的法院还可以对他们再审理吗？

本篇为神奇组织之1972苏联粮食大劫案。