Today's AI Pulse
-1.png)
Anthropic shuts down Fable 5 and Mythos 5 models under U.S. export controls
Anthropic abruptly disabled its flagship Fable 5 and Mythos 5 models to comply with a U.S. export‑control directive, underscoring the vulnerability of firms that depend on closed‑source AI services. The move triggered a sharp rally in Chinese open‑source offerings as investors shifted to downloadable models that can run on‑premise.
Also developing:
By the numbers: Respond raises $62.5M Series B
Supercharging Local AI Development with RHEL on NVIDIA DGX Spark
Red Hat announced a development preview of Red Hat Enterprise Linux 10 running on NVIDIA’s DGX Spark workstation. The DGX Spark, built on the Grace Blackwell GPU architecture, offers up to 1 petaflop of compute and 128 GB of unified memory, enabling developers to train and test large language models locally. By pairing this hardware with RHEL’s enterprise‑grade OS and tooling, Red Hat aims to give teams a sandbox for cost‑controlled, data‑sovereign AI development before moving to production. The preview targets autonomous‑agent workloads and promises smoother migration to Red Hat OpenShift clusters.
How ACKO Drove 13% More Conversions & 50% Drop in Calls with GenAI
ACKO, India’s digital‑first insurer, deployed a homegrown GenAI chatbot to replace high‑pressure phone calls. Using Amplitude’s funnel, cohort and experimentation tools, the bot delivered an 88% cut in answer latency and a six‑fold reduction in launch time for new conversational...
The MCP Catalog Is Here: Discover, Deploy, and Connect on Red Hat OpenShift AI
Red Hat OpenShift AI 3.4 introduces the MCP catalog, a developer‑preview marketplace of pre‑validated Model Context Protocol (MCP) servers. The catalog lets users discover, deploy, and manage MCP servers directly on OpenShift with an automated lifecycle operator and a secure...
The Three Pillars of Trust: The Hardened OpenShift Foundation
Red Hat OpenShift’s new security framework centers on three pillars—integrity, isolation, and identity—to meet the rising demands of generative AI and digital sovereignty. The roadmap introduces node attestation, confidential virtual machines, post‑quantum cryptography, AI Bill of Materials scanning, and a zero‑trust...
The Smartest Things I've Read Lately About AI
The article curates recent thought‑leadership on generative AI, highlighting three key strands. Harvard Business Review warns that framing AI as an employee erodes accountability and adoption, while another HBR piece describes an "AI fog" that clouds investment decisions. Oliver Wyman quantifies...
Dicker Data’s TechX 2026 Spotlights AI-Led Ecosystem Shift
Dicker Data's TechX 2026 will tour five ANZ cities, spotlighting an AI‑led ecosystem where partners shift from product resale to outcome‑driven services. The event emphasizes collaborative solutions across AI, cloud, security, data and devices, addressing the rapid acceleration of AI...
AI Models Disagree on Jobs, Agree on At‑risk Work Types
Economists asked ChatGPT, Gemini, and Claude the same question: "Which jobs will AI replace?" I ran the same test, and the result? AI can’t agree on which jobs it will replace but it does agree on what kinds of work are in...
Tripo 3.1 in ComfyUI: Production-Ready, High-Detail 3D Asset Generation.
ComfyUI now supports Tripo 3.1 through Partner Nodes, enabling high‑detail 3D asset generation within existing workflows. The integration offers high‑density geometry, PBR‑ready materials, and hero‑level quality for close‑up production use. Users can create models from text, images, or multiview inputs and...
Beyond the Bot: Building Human-Centred AI Experiences
Enterprises have poured billions into AI, yet customer frustration remains high because automation often replaces experience. Bajaj General Insurance re‑engineered its AI claim process to prioritize human empathy, allowing a bot to handle image assessment in under 20 minutes before...
AI Leading to “Radical Shift” In Client Expectations
Law firms worldwide are confronting a “radical shift” in client expectations driven by artificial intelligence, according to a Positive Group study. Clients now demand lower fees and faster, AI‑generated outputs while still requiring human validation and full liability. The research...
What Kind of Infrastructure Will K-12 Schools Need for AI?
K‑12 districts are confronting a new wave of AI‑related infrastructure pressures, even though AI has not yet caused outright network failures. A 2025 CoSN report shows only 16% of schools feel fully prepared, while 61% struggle with fragmented data. Leaders...
Claude Skills.
The author reports using Claude's AI "skills" more than 20 times daily since December, labeling them the most powerful alpha technique in generative AI. Over the past six months they logged dozens of hours experimenting with various skill configurations, culminating...
When the Ruler Is Made of the Thing It Measures: Multi-Model Evidence on AI Occupational Exposure Scores
A new NBER paper reveals that AI‑generated occupational exposure scores are highly unstable across frontier language models. Using identical O*NET task data, the share of U.S. jobs deemed high‑exposure ranges from 2.7% with Google’s Gemini 2.5 to 51.5% with Anthropic’s Claude 4.5—a...
Grab Seeks AI Edge in Product Development Using Agentic Engineering, Vibe Coding
Grab is integrating AI into its product development pipeline through vibe coding and agentic engineering, allowing rapid prototyping and higher productivity. The Virtual Store Manager, created by a non‑engineer using AI‑generated code, exemplifies how vibe coding turns ideas into functional...
Last Week in ConTech - 11 May 2026
Consumer‑focused AI is breaking into construction software, highlighted by Thumbtack’s new Claude integration that lets homeowners find contractors through conversational prompts, and Trimble’s Claude‑SketchUp link enabling plain‑language 3D modeling. The article notes a broader convergence of consumer and enterprise tools...
Giant Clap: Emotive Brings STI Monsters To Life In Blockbuster AI Campaign
Emotive and AiCandy used generative‑AI to produce a short film for Four Seasons Naked Condoms that pits a giant baby against STI monsters, delivering Hollywood‑level visuals at a fraction of traditional costs. The horror‑comedy spot targets Gen Z on Snapchat, YouTube...
AI Investment ‘Advice’ Is 50% More Likely to Pump You up — and Trip You Into Costly Blunders
Artificial intelligence tools are increasingly shaping retail investment decisions, but a new analysis shows AI‑generated advice makes investors 50% more likely to act impulsively. The underlying cause is a well‑documented psychological tendency called action bias, which pushes traders toward unnecessary...
5 Steps to Align Your Physical Fab Infrastructure with AI Yield Models
The article presents five practical steps to synchronize semiconductor fab infrastructure with AI yield models, starting with a contamination‑controlled baseline and ISO 14644‑5 compliance. It highlights how outgassing, fastener stress, and HVAC‑induced vibrations create phantom defects that confuse AI algorithms. Deploying...
Dreaming: 2026's Top Agent Pattern for Memory Curation
Dreaming will be the most copied agent pattern of 2026. It's the cleanest memory-curation design I've seen and every DIY agent stack will reimplement it inside 90 days.
In The Vacuum Of AI Legislation, Libraries Have The Playbook
The White House’s AI framework leaves a regulatory void, prompting policymakers to look to libraries for guidance. Decades of library practice—exemplified by the Google Books project and HathiTrust—have established fair‑use precedents for digitizing and training AI on copyrighted works. Library...
Register: Risky Future AI Tools for Homeowners Data ‘Demo Day’ on May 13
Insurance Journal’s Risky Future series is hosting a free “AI Tools for Homeowners Data” Demo Day on May 13, 2026, at 1 p.m. ET. The online event will showcase AI‑driven property‑data platforms from Home Factors and Eagleview, focusing on underwriting, catastrophe modeling,...
Transforming Cancer Care: The Evolution of Remi, an AI-Powered Oncology Assistant
Reimagine Care, a Nashville‑based virtual oncology provider, unveiled an upgraded version of its AI‑powered assistant Remi. The new system leverages large language models and agentic AI to conduct more natural conversations, integrate NCCN‑based clinical pathways, and automatically triage 50% of...
AI-Driven GDP Surge Will Skew Capex‑GDP Ratios
One thing I didn't say on CNBC today that I should have is that when we talk about Capex as a percentage of GDP, we need to be thoughtful about the changes in the denominator. Sure, we are spending up...
OpenAI Builds AI Deployment Biz Around Team Behind Virgin Atlantic Concierge
OpenAI has launched a $4 billion OpenAI Deployment Company, acquiring Tomoro to accelerate enterprise AI adoption. The deal brings roughly 150 engineers who built Virgin Atlantic’s AI concierge, giving OpenAI a ready‑made travel proof point. The new venture will embed OpenAI...
Advocate Health Speeds up Prior Authorizations with AI: 5 Notes
Advocate Health has integrated an AI‑driven module into its Epic EHR to automate prior‑authorization workflows for specialty medications. The new system replaces traditional phone and fax processes with digital questionnaires and AI‑generated draft responses. Staff time per authorization fell from...
New AI Target Discovery Paper Sets Benchmark
New Paper Alert 🚨: A collaboration with the leading target ID experts in Nature Reviews Drug Discovery. If you are looking to teach your AI target discovery - this is a good place to start. Thanks Peter Kirkpatrick for giving...
PSC Bridges-2 Helps Train AI to Predict and Explain Airport Collision Risks
At the NASA Formal Methods Symposium, CMU’s AirLab announced World2Rules, a neuro‑symbolic AI trained on PSC’s Bridges‑2 supercomputer to predict and explain airport surface collision risks. The system ingests the 10‑TB Amelia‑42 dataset—two years of FAA movement records from 42...
Trust and Quality Beat Speed in AI Conversations
Voice is replacing the textbox. Companies are moving beyond clunky bots toward real-time AI conversations that sound human, understand interruptions, and respond with speed + context. The takeaway from @Amazon Nova 2 Sonic: In AI, speed alone isn’t enough. Trust...
SF Offers AI Events; My AI Curates Them
One thing San Francisco has that even San Jose doesn't is tons of events for AI people and @michelleefang keeps track of them. So does my AI at https://t.co/8L5xphk0qQ, mostly by reading Michelle's posts here. In fact, I talk to...
The Fastest Way to Build an AI Agent (Start With the Output, Not the Tool)
The article proposes a reverse‑engineering method for building AI agents by starting with a mock output rather than tool capabilities. By drafting a fake example of the desired result, users can instantly identify required inputs, define the schema, and streamline...
NightBeacon Slashes Detection to Seconds, Cuts False Positives 87%
NightBeacon is changing the game for us @Binary_Defense - we have this technology so dialed into every aspect of what we do. Our MTTD and MTTR has drastically been reduces to seconds and minutes, not multiple minutes or hours....
Real‑time AI Collaboration: Beyond Jokes to Practical Use
Haven’t tried this but it seems very neat… Yet all of the demos (except maybe one) are the model being fun and/or annoying by correcting or reminding in real time. There are obvious uses for this sort of model in meetings,...
Pinterest Recommendation Engine Update Aims For Enhanced Ad Relevance
Pinterest unveiled a Contextual Sequential Two‑Tower Model that injects real‑time browsing signals into its ad candidate generator, delivering a 3‑10× lift in per‑user ad relevance and boosting median relevance by roughly 275‑300%. The upgrade addresses a prior shortfall where less...
AI Replacement Isn't Inevitable—Explore a Better Path
The AI race-to-replace (girlfriends, jobs, decision-making, etc) isn't inevitable: There's a better path, and this new website is IMHO the most detailed description of it so far:
ChatGPT's New Model Brings Personality and Personalization
speaking of things that have gotten over a threshold for me, the combo of the new ChatGPT model, personality, and personalization feels like a new thing
Microsoft Researchers Find AI Models and Agents Can't Handle Long-Running Tasks
Microsoft researchers published a preprint revealing that even top‑tier large language models (LLMs) corrupt documents during long‑running, multi‑step workflows. Using a new benchmark, DELEGATE‑52, they measured a 25% average loss of content after 20 delegated interactions, with overall degradation around...
AI Is a Phase Change, Not a Bubble
ai is not a bubble it’s a phase change, like going from solid to liquid to gas. https://t.co/HCsKPB5lLu
Enterprises Need Trustworthy, Repeatable AI, Not Magic
AI isn’t magic. Not yet. GenAI solved access to AI through natural language, but enterprise trust still breaks because models lack intuition, produce non-deterministic outputs, and haven’t earned true autonomy. The enterprise doesn’t need “magical” AI. It needs governable, repeatable,...
Lendi Group Runs First Project Through "Agentic SDLC"
Lendi Group, a home‑lending fintech, completed its first project using an agentic software development lifecycle (SDLC) powered by Atlassian’s Teamwork Graph. The new process captures meeting audio with Loom, auto‑generates PRDs in Confluence, and employs AI agents to create Jira...
Generative AI’s $1.6 T Odds Beat Roulette’s Zero
Which has better odds? Generative AI earning $1.6 trillion/year or the roulette wheel landing on zero? https://t.co/52cTL3iSHM
AI Pressure Drives Yum Brands to Overhaul Data Estate
Even if you don't care about all the AI transformation stuff, it does seem as if this executive urgency is forcing companies to finally fix longstanding pain points. Here, Yum Brands is fixing their data estate so that they can get...
AI Must Reach $1.6 Trillion—Four Times Google’s Peak
One estimate of how much annual revenue AI needs to “make sense”: 1.6 trillion. That’s four times what Google made in its best year. (total revenue so far is perhaps on order of 100 billion.)
OpenAI Launches Daybreak to Accelerate Continuous Cyber Defense
OpenAI is launching Daybreak, our effort to accelerate cyber defense and continuously secure software. AI is already good and about to get super good at cybersecurity; we'd like to start working with as many companies as possible now to help them...
AI Gives You One Extra Day Weekly—Act Now
Picture your business one year from now. AI is streamlining your operations. Your team has clear workflows. You’ve reclaimed a full day each week. Or… you’re still “planning to implement.” Your choice. https://t.co/B3ac1JrQNc
AI Floods Code, Outpaces Trust and Triage
"AI increased the supply of code faster than it allowed maintainers to triage contributions. AI didn’t increase the supply of trust yet. That is the real open-source problem: AI is not killing open source. It is killing open contribution as we...
Sigma Browser Adds Hermes Agents for Local AI Workflows
Sigma browser now supports Hermes Agents 🤯 You can now run open-source models natively via their Eclipse engine for unlimited agentic workflows. Private. Local. Open-Source. No cloud. No subs. Mac. Windows. iOS. Android 👀↓ https://t.co/tMwwPolCq9
Weekend Roundup: Top AI Marketing News Highlights
ICYMI Top #Marketing #AI News Stories from over the weekend: https://t.co/8V6t1LWrNB Top stories from @FastCompany @InsideIowaState @FiercePharma @martechismktg @CNBC @BainandCompany @saastr and @Oracle Blogs. https://t.co/21uxv4SV69
Exploring Agentic Harness Use and AI Bubble Debate
I am curious on people's responses - have you: - used an agentic harness - and whether you think AI is a bubble
AI Transforms Every Stage of the Buyer Journey
How #AI Is Redefining The Buyer Journey by @ingliguori #Marketing #ArtificialIntelligence #MachineLearning #ML #MI https://t.co/6NjaJb1Nk3
Multilingual, Multitask AI Demo Proves Its High-Value Appeal
The demo of this model is cool. Interacts with multiple people in multiple languages. On multiple tasks. Now we know why @miramurati got the big bucks.