Today's AI Pulse
Odyssey lands $310M Series B, valued at $1.45B
Odyssey announced a $310 million Series B round that values the AI lab at $1.45 billion. The round was led by Natural Capital and included backers such as Amazon, AMD Ventures, GV, EQT and IQT. In parallel, Odyssey struck a preferred‑cloud deal with AWS to use Trainium silicon for large‑scale model training.
Also developing:
Hippocratic AI Expands Footprint, Partners with Leading Health Systems to Improve Patient Outcomes
Hippocratic AI announced new collaborations with Cincinnati Children’s, UNC Health and the Gift of Life Marrow Registry, extending its safety‑first generative AI tools across pediatrics, primary‑care scheduling and donor outreach. The company’s AI Front Door and Nurse Co‑Pilot are now live at Cincinnati Children’s, while UNC Health will use AI to streamline appointment booking and specialty pharmacy calls. Gift of Life will employ AI‑driven voice outreach for its 550,000‑plus donor registry. To date Hippocratic AI’s Polaris Constellation Architecture has powered more than 180 million patient interactions with a 99.90% clinical accuracy rate and zero severe harm incidents.
Instacart Teams Up with Claude to Power AI‑Driven Grocery Cart Creation
Instacart has integrated its grocery platform with Anthropic's Claude conversational AI, allowing U.S. users to assemble shopping carts directly within a chat interface. The feature taps Instacart’s real‑time inventory across more than 2,200 retail banners and a catalog of over...
Google Retires Dynamic Search Ads, Makes AI Max Default for Search Campaigns
Google announced it will retire Dynamic Search Ads and require advertisers to migrate to its AI Max for Search platform, with automatic upgrades beginning in September. The move consolidates targeting, creative and optimization into a single AI‑driven system that early...
Jeff Bezos' Project Prometheus Targets $10B Funding as Cursor Pursues $2B Round
Jeff Bezos is finalizing a $10 billion financing round for Project Prometheus, which could lift its valuation to $38 billion. At the same time, AI coding platform Cursor is in talks to raise $2 billion, targeting a valuation north of $50 billion. Both deals highlight...
Infor Rolls Out AI Orchestration Suite to Accelerate Enterprise AI Scaling
Infor announced the launch of its Agentic Orchestrator and an expanded Velocity Suite, tools designed to speed AI deployment across large organizations. Backed by research showing half of enterprises are still in pilot phases, the suite promises measurable gains such...
Indirect Prompt‑Injection Attacks Emerge as Top Threat to Enterprise LLMs
Security researchers have flagged indirect prompt‑injection attacks as the leading vulnerability for large language models (LLMs). The rise of enterprise AI tools such as Zeta Global’s Athena, built on OpenAI technology, amplifies the urgency for robust mitigation strategies.
Coordinating AI Agents: The Next Hard Frontier
Organizational design for agents is hard, benchmarking agents working in concert is hard. Together, this is the next critical frontier for making AI matter in economically valuable tasks, and we really don’t know very much about it.
DOJ Teams with Elon Musk’s xAI to Contest Colorado AI Bias Law
The U.S. Department of Justice filed an amicus brief backing Elon Musk’s xAI in a federal lawsuit challenging Colorado’s AI discrimination law. The DOJ argues the statute violates the Equal Protection Clause and could undermine America’s AI competitiveness. The case,...
The AI Playbook That Built an $80M 1-Person Business (You’re 1 Prompt Away and Don’t Know It)
A solo founder built an AI platform that sold for roughly $80 million in under six months, demonstrating how a one‑person operation can achieve an eight‑figure exit. The story underscores that the biggest AI wins come from eliminating barriers that keep...
Extra Claude Hours Fuel Dashboard Update
Claude's recent weekly limit reset gave me an extra 8 hours and an extra Claude Design quota, so I used it update my session metrics skill plugin HTML dashboards https://ai.georgeliu.com/p/my-claude-code-plugin-marketplace 😎
Scaling AI Tech Without People Creates Operational Debt
THE Operational Debt of AI Startups Praktika .ai: Great Tech, Failing Infrastructure. As someone who keeps an eye on digital trends and engagement, I’m always looking for tools that actually move the needle. For a while, @praktika.ai was that tool. The...
Get Started with Claude Design Today
Anthropic has unveiled Claude Design, an AI‑driven interface‑building tool that lets product managers generate UI mockups from simple text prompts. The launch includes a publicly available GitHub repository and a curated list of top use cases to accelerate adoption. Anthropic...
Osaic Adds Scribe AI to Craft Step-by-Step Guides for Advisors
Osaic, the platform serving more than 10,000 financial advisors, has integrated Scribe, an AI‑driven workflow‑documentation tool. Scribe automatically records screen actions and converts them into step‑by‑step guides, promising documentation up to 12× faster and information retrieval 63% quicker. The partnership...
Vietnam, China News Agencies Partner on Digital Media, AI
Vietnam News Agency (VNA) and China’s Xinhua News Agency signed a deeper cooperation pact focusing on digital transformation, AI‑driven journalism, and data exchange. The agreement, concluded during a VNA delegation visit to Beijing, includes joint development of AI content generation,...
Hong Kong: PolyU AI Laboratory to Advance On-Device Intelligence
The Hong Kong Polytechnic University (PolyU) has entered a strategic partnership to launch a joint laboratory dedicated to on‑device artificial intelligence. The lab will focus on low‑precision training, inference and model‑fusion technologies that run locally on hardware, reducing reliance on...
Utah Medical Board’s Call To Halt AI Drug Pilot Triggers Med Practice Showdown
Utah’s medical board has ordered an immediate pause on the state’s pioneering AI‑driven prescription renewal pilot, the first of its kind in the nation. The pilot, launched by a consortium of health systems and a tech vendor, used machine‑learning algorithms...
Co‑evolving Robot Bodies and Brains via Leader‑Follower Game
ICLR 2026: This paper uses a "Leader-Follower" game to design robots. The Leader evolves the robot's body (e.g. adding limbs & joints) while predicting how the Follower (the brain) will learn to move it. This teamwork creates specialized robots—like creatures...
Big Four Accounting Chooses AI over Humans, Cuts Benefits & Hiring
The Big Four accounting firms are accelerating AI adoption while slashing staff and benefits. KPMG recently eliminated 10% of its U.S. audit partners, and Deloitte cut paid time off, froze its pension plan, and halved family‑leave benefits. Junior hiring has...
Meta’s Compute Grab Continues with Agreement to Deploy Tens of Millions of AWS Graviton Cores
Meta announced a partnership with Amazon Web Services to add tens of millions of Graviton5 CPU cores—each chip packing 192 cores—to its AI compute portfolio, bolstering the Llama builder platform. The agreement expands Meta’s already diverse hardware ecosystem that includes...
ARC‑AGI‑2 Shows LOW Mode Outperforms Claims
To everyone preaching “GPT-5.5 on LOW reasoning”… How do you explain these ARC-AGI-2 results? Low gets absolutely cooked. Genuinely curious what you are actually using it for? Code? UX? (I’m still riding xHigh all day) But maybe I'm missing something. CCing folks I saw...
AI‑empowered Patients Demand Smarter Clinician Collaboration
Patients are empowered by AI in a way they've never been before. This raises the bar for clinicians everywhere. I'm hearing from clinicians that their patients now come to clinic having done tremendous amounts of research on themselves and their conditions...
Responsibility Is the Human Moat
The authors argue that only natural persons can bear legal responsibility, creating a "human moat" that AI cannot cross. They outline a reference document defining responsible work, emphasizing that every business service must be overseen by a competent individual. AI...
Spent $40k Building 70 AI Agents to Disrupt Finance
I have used over 4 million credits on Perplexity. Each credit is one penny. In 8 weeks. Ive built 70 AI agents. I’m about to disrupt the entire financial services industry for fund issuers and financial advisors. I’m about to remind you all...
Everything Announced At Google Cloud Next In Under 15 Minutes
Google Cloud Next’s rapid‑fire keynote, led by CEO Thomas Kurian, showcased a slate of AI‑centric announcements. The company introduced new generative AI tools, including Vertex AI Studio and an expanded AI Hub, while unveiling upgraded TPU v5e hardware. Partnerships with...
How Eye Contact Shapes the Believability of Computer-Generated Faces
Researchers at the University of Western Australia examined how eye‑gaze direction influences the perceived authenticity of computer‑generated faces. In experiments with 150 participants, avatars showing anger or happiness were judged most believable when maintaining direct eye contact, while sad avatars...
Chinese Open‑Source AI Powers Major Apps, Underrated Advantage
Everyone thinks OpenAI and Google are winning the AI war. They are wrong. The apps you use every day are running on Chinese open-source AI. And almost nobody's talking about it. Cursor's code editor runs on Kimi K2.5. Shopify switched to Alibaba's Qwen...
Inside the Navy’s Push for Resilient, AI-Enabled Command Centers
The Navy is overhauling its maritime operations centers (MOCs) with resilient cloud, on‑prem compute, modular technology and AI‑driven orchestration. Officials highlighted decentralized digital architectures that keep data and applications available even when satellite links fail. AI will automate data aggregation,...
Patent Analysis Is Increasingly Shaping AI-Driven Target and Drug Candidate Selection
A new review in Nature Reviews Drug Discovery outlines how AI is expanding target identification by merging multi‑omics, knowledge graphs, and foundation models. The authors argue that patentability, commercial tractability, and competitor analysis should be evaluated alongside druggability and safety...
DeepSeek's New Models Are so Efficient They'll Run on a Toaster ... By Which We Mean Huawei's NPUs
Chinese AI firm DeepSeek has launched V4, an open‑weights large language model family that includes a 284 billion‑parameter Flash MoE variant and a 1.6 trillion‑parameter Pro version. The models employ a hybrid attention mechanism and FP8/FP4 quantization, delivering up to 13.7× lower...
Hawley Urges GOP to Reject AI Money, Warns Populist Backlash
Senator Josh Hawley is encouraging Republicans to shun the $300M in lobbying money from the AI industry and instead regulate the industry to avoid the ire of their voters. AI is now viewed so negatively that I can see a Trump-like...
DeepSeek Drops Cheaper V4 AI as Huawei Jumps In
DeepSeek unveiled its V4 AI model family, featuring a 1.6‑trillion‑parameter V4‑Pro and a 284‑billion‑parameter V4‑Flash, both supporting up to one million‑token context windows. Pricing is dramatically lower, at $3.48 and $0.28 per million output tokens respectively, undercutting OpenAI and Anthropic rates....
B*tchwork My AI Did for Me - Part 7: Found 20 PRISM Leads on Instagram and Sent Approved DMs
In part seven of her "B*tchwork" series, Liz Elliott shows how she leveraged AI to mine Instagram for 20 PRISM‑qualified leads and automatically dispatch pre‑approved direct messages. She frames AI as a middle‑man tool that handles sorting, checking, and first‑pass...
Facebook News Creator Shares AI-Generated Image of Body Bags at Hastings Triple-Homicide - Police and Netsafe Issue Warning over Fake...
A Facebook news page shared an AI‑generated image depicting body bags at the Hastings triple‑homicide scene, prompting criticism from police and online‑safety group Netsafe. The image, posted by Australia/NZ Crime TV, contained obvious AI errors such as a misspelled “ambulance”...
ChatGPT's Latest Update Makes It Harder Than Ever to Spot AI-Generated Images
OpenAI unveiled Images 2.0, the latest ChatGPT image model that can generate up to eight highly realistic images per prompt and render convincing text, including handwritten notes and news headlines. The model uses a "thinking" step to break down requests, improving...
#516 - Amazon Japan, AI Ads, And Global Growth
In this episode, host Bradley Sutton chats with longtime Japan resident and e‑commerce expert Martin about the unique dynamics of selling on Amazon Japan and the broader Japanese marketplace. Martin explains how Amazon Japan has become the third‑largest Amazon marketplace...
AI Threatens Trust in Doctors, Especially Among Youth
AI is going to radically reshape who we trust and how we get information. A new report from @EdelmanPR finds that AI is reshaping perceptions of medical expertise, with 64% of consumers believing users fluent with AI can match or...
Switching LLMs in LangChain Boosts Apps Tenfold
One of the most magical moments is when you update your LangChain harness to upgrade the LLM model and your entire application gets 10X better. Feels like the old days when you upgraded the CPU from 486 to Pentium on your...
OpenAI Codex Brings Agent AI Workflows to Coding Tasks
OpenAI unveiled Codex, an AI‑powered coding agent that can generate, execute, and edit code across multiple languages while running tasks in parallel cloud sandboxes. The platform now ships with the GPT‑5.5 model, positioning it as the default engine for complex...
Parallel AI Agents and Result Aggregation Boost Offensive Success
Sometimes success of using AI agents for offense is using them in multiple or parallel rounds. With different models. And aggregating the results.
AI Conjures Silicon Valley‑styled AAA Game Screenshots
"chatgpt, generate screenshots from a AAA video game themed around trying to make it in silicon valley" lol - gpt Image 2 is amazing also, weirdly, I want to play this?? https://t.co/1UX5LhqyeY
DeepSeek Open-Sources V4 Large Language Model Series
DeepSeek unveiled its open‑source V4 large language model series, featuring the flagship V4‑Pro with 1.6 trillion parameters (activating 49 billion per prompt) and the lighter V4‑Flash with 284 billion parameters (activating 13 billion). Both models use a mixture‑of‑experts architecture and introduce a hybrid attention...
Cursor's SpaceX, Anthropic Deal Highlights Massive Compute Costs
Behind Cursor’s Deal With SpaceX, Anthropic and Compute Costs Loomed Large — The Information https://t.co/pfTwNxybKN
Cybertruck Feels Humanlike with 14.3.2 Update
It's honestly kind of freaky how humanlike my Cybertruck's driving is with 14.3.2 Have a great weekend everyone. Go put on some good music and self-drive somewhere fun. Comment with where you're self-driving this weekend. https://t.co/8TxFC71g45
Managing Thousands of AI Agents Is the Real Challenge
Check out the latest article in my newsletter: Managing a Fleet of Thousands of AI Agents Is the Real Problem. @GoogleAI Showed Its Answer. https://t.co/mXDBV2eoGR via @LinkedIn
Honolulu Airport's AI Music Welcomes Tourists, Sounds Awful
Honolulu airport is welcoming tourists using all genAI music and it sounds like absolute trash. https://t.co/YBUzyAJHJl
AI Hardware War: Nvidia, Google, Amazon, Huawei Dominate
The fighters in the battle for the future of AI: - NVIDIA (heavyweight champion) - Google (TPUs) - Amazon (Graviton) - Huawei (Ascend) And then a bunch of startups with neuromorphic or mega-chip strategies that will, if they look promising, get bought up. Anything else?...
AI Resistance Is Growing; Time Is Running Out
Most of You are Resisting AI. The Data Says You’re Running Out of Time by @Timothy_Hughes https://t.co/49cwgynMOj @DLAIgnite #SocialSelling #DigitalSelling #Sales #SalesTips #SalesLeader #ArtificialIntelligence #Technology #Innovation #Tech #TechNews #Marketing #Strategy https://t.co/cI5gFCGuWK
Your AI Skill, Not Your Time, Becomes Currency
In the near future, Everyone will have their own AI Skill. They'll sell that instead of selling their time.
Silicon Valley AI Agents Waste Tokens, Create Chaos
SiIicon Valley’s #AIAgent hiccups: Wasted tokens and ‘chaotic’ systems by Jonathan Vanian @cnbc Learn more: https://t.co/J9FHR3RSEl #AI #GenerativeAI #ArtificialIntelligence #MachineLearning #ML https://t.co/AGhCIaOm9w
Use AI to Reorganize Existing Work, Not Rebuild
🔑 The real shift: stop thinking "I need to build this from scratch." 💡 Start thinking "I already have something, how can AI help me reorganize it?" 📝 New blog post with prompts to help you do exactly that. https://t.co/s9XnH6TjMf