AI-powered podcast about AI news. Aggregates a number of RSS feeds and uses an llm pipeline to select and double check stories.
Today's briefing: Google's Gemini Omni Flash adds video editing, Anthropic documents its Claude Advisor tool, and China's new rules prompt ByteDance and Alibaba to disable humanlike AI agents.
Today's AI news: Meta's Muse Spark Watermelon update, a record surge in AI-found vulnerabilities, Bridgewater's proprietary finance model, Anthropic's China access crackdown, Alibaba's Claude ban, and a tip for Deepseek's DSpark.
Meta's Watermelon AI model reportedly matches GPT-5.5, AWS raises AI agent quotas, Microsoft invests $2.5B in enterprise AI, and companies throttle AI use due to cost.
Today's AI briefing: The Trump administration lifts restrictions on Anthropic's Claude models. AWS claims 70% storage cost reduction for AI logs. Plus, preventing agent bloat, securing AI agents, Meta's cloud business, and questioning model benchmarks.
Today's AI briefing: Anthropic's Claude Sonnet 5 brings agentic AI to a lower price tier. Microsoft's Memora research cuts agent memory use by 98%. A new study finds heavy AI adoption is linked to more hiring, not layoffs. Plus, a new AI browser attack, China's trillion-parameter model, and OpenAI's safety standards foundation.
Explore Shopify's public Slack AI coding agent, OpenAI's new standards alliance, Amazon's cost‑saving model shift, and open‑source security coalitions.
Google limits Meta's Gemini access as AI demand strains cloud capacity. Coinbase cuts AI spending 50% by switching to Chinese models. Nvidia partners on 360MW Indonesia data center. Z.ai's GLM-5.2 matches US models at finding security bugs. Open-weight models now run from 1.6T to 230M parameters on Raspberry Pi.
Today's Compact Conversations covers Google's integration of native computer control in Gemini 3.5 Flash, OpenAI's new Jalapeño AI chip, and practical tips for building reliable AI agents.
Anthropic’s Mythos model faces a breach claim and export ban, while five Chinese AI labs slash token prices up to 99%.
Compact Conversations for June 20, 2026: a Claude Code plugin quizzes developers on AI-written code before they move on. Plus OpenAI's $3.7 billion burn, enterprise AI cost controls, data center power fast-tracking, benchmark gaps, Claude Fable 5 testing issues, and Nvidia's Nemotron 3 Ultra.
Microsoft patches a critical M365 Copilot flaw, SpaceX moves to buy Cursor, Anthropic walks back Claude Agent SDK billing, Nvidia bets on local AI PCs, and a CIO Dive study finds AI value trapped in pilot purgatory.