Good morning, welcome to this week’s Lore Brief, your 3-minute brief of the most important moves in AI and tech.
This issue is brought to you by Factory, the agent-native platform for building, testing, and deploying software with AI coding agents.
This week is Google I/O, and the company dropped a flood of major AI updates across models, agents, video, and developer tools. From world-understanding video models to always-on personal agents and a mobile AI Studio, Google is moving fast to close the gap in the frontier race.
Google DeepMind launches Gemini Omni → The new model family can generate and edit video from any input with real physics understanding, consistent characters, and logical storytelling. Gemini Omni Flash is live now in the Gemini app, Flow, and YouTube Shorts, with full API support coming soon. Read more here →
Andrej Karpathy joins Anthropic → The legendary AI researcher and educator announced he’s heading to Anthropic for frontier LLM R&D work. The news went mega-viral on X as one of the biggest talent moves of the year. Read more here →
Google releases Gemini 3.5 Flash → The new model beats 3.1 Pro on coding and agentic benchmarks while running 4x faster than other frontier models and up to 12x faster inside Antigravity. It’s available now in the Gemini app, AI Studio, and more, with the Pro version coming soon. Read more here →
ChatGPT solved a famous open math problem:
Gemini Omni demo by Demis:
OpenAI offers to invest $2M in tokens in every YC startup:
Cursor ships Composer 2.5 and teams with xAI → The upgraded coding model is smarter on long tasks and up to 10x more efficient. Cursor is now training an even larger model from scratch using xAI’s Colossus supercluster. Read more here →
Google introduces Gemini Spark → This 24/7 personal AI agent runs in the background on dedicated Google Cloud VMs so it keeps working even when your devices are off. Built on Gemini 3.5 and Antigravity, it handles long tasks and will soon connect to third-party tools via MCP. Read more here →
Dan McAteer predicts memory breakthrough in 2026 → The agentic AI engineer says this is the year we finally crack persistent memory, leading to infinite context and true continual learning. The post struck a chord with the builder community. Read more here →
GoodfireAI hits $1.25 billion valuation → The interpretability startup is building tools to peer inside AI models and make them safer and more controllable. Their work was featured in Forbes as one of the most important problems in AI right now. Read more here →
Gmail rolls out AI Inbox → The updated feature surfaces ready-to-send drafts, links to your Docs and Sheets, and one-click task controls to help you clear clutter fast. It’s live for AI Plus and Pro subscribers. Read more here →
Google AI Studio goes mobile → The new iOS and Android app lets you build prototypes and full apps straight from prompts wherever you are. It’s coming soon to the app stores. Read more here →
