Three daily shows from the alley — AI news at dawn, gaming at lunch, one deep-dive after dark. Every day, Monterrey time.
The day's AI news before your coffee cools. Daily from The Penguin Alley.
Listen →
The day's gaming news and what AI is doing to games. Daily from The Penguin Alley.
Listen →
One topic, told properly. Evening deep-dives from The Penguin Alley.
Listen →The AI Engineer World's Fair ended not with applause but with a fight about loops — we go under the headline to explain what agentic loops actually are, why the field is split on them, and what that means for what you build next.
The Steam Summer Sale is live with real deals on 2026's best releases — we dig into what's actually worth your money. Plus: Slay the Spire 2 adds 15 co-op-specific cards and makes first-time modding way less painful, and Sandfall's director says 'maybe people won't like' their next game, which is either the most refreshing thing a studio head has said all year or a promise we'll hold them to.
Google DeepMind and indie film powerhouse A24 just announced a research partnership nobody saw coming — plus the loops debate that ended the AI Engineer World's Fair, a practical Fable tip from the conference floor, and a new CLI that keeps your AI agent from shipping vulnerable packages.
A leading open-source game engine slams the door on almost all AI-generated code, and we get into why that matters for the people who actually build games. Then: the Subnautica 2 lawsuit finally settles, Sony inches toward an all-digital PlayStation, and Notch calls the ESA 'borderline evil' over private servers.
Google DeepMind ships two new models builders can use today — Nano Banana 2 Lite and Gemini Omni Flash — while Anthropic's new Claude Science workbench pulls AI into the lab. Plus how OpenAI engineers used core-dump forensics to kill an 18-year-old bug, and a lightning round of fresh benchmarks.
AI agents are increasingly "improved" without any retraining — by evolving their own reflections, playbooks, and cheat-sheets while the model underneath stays frozen. Tonight we go under the headline to ask whether that self-evolution is real progress or just a win on the one benchmark that flattered it, and how held-out selection tries to tell the difference.
Microsoft insists it isn't pulling back on games even as reports point to Arkane's closure and layoffs at Hitman maker IO Interactive over its scrapped Xbox-partnered RPG. We dig into the gap between the spin and the cuts, and close on a bittersweet send-off for a Half-Life 2 legend.
Anthropic's Claude models hit general availability in Microsoft Foundry on NVIDIA's GB300 Blackwell Ultra GPUs, and we unpack what it actually means for Azure-native builders. Plus governing autonomous agents, NVIDIA Jetson in lunar orbit, and a fast research lightning round.
A single day of AI API calls outran a month of server bills. We dive under the headline into retry storms — the old distributed-systems failure that token-metered pricing just made expensive again — and the backoff, jitter, and idempotency habits that tame it.
A reported billion dollars in digital GTA 6 preorders has GameStop and physical retail sweating. Plus: unionized Xbox workers brace for layoffs, third-party Game Pass deals go on pause, and Dbrand's Companion Cube case meets Valve's legal incinerator.
A special from opening day of the AI Engineer World's Fair in San Francisco: why open-weight models like NVIDIA's Nemotron are moving into serious production with Palantir and U.S. agencies, plus OpenAI's new map of how AI could reshape EU jobs — and what 6,000 engineers should watch this week.
We dive under tonight's top Hacker News story — DeepSeek's DSpark paper — to explain speculative decoding from first principles: how a small draft model lets a big model generate text faster without changing a single word of the output.