Today's AI Pulse
OpenAI launches ChatGPT Images 2.0 with multilingual text and agentic reasoning
OpenAI has introduced ChatGPT Images 2.0, a next‑generation image model that adds multilingual typography, real‑time web research and agentic reasoning. The new “Thinking” mode lets paid users generate up to eight coherent images per prompt and create floor plans, maps and UI mock‑ups. The rollout expands OpenAI’s generative‑AI capabilities beyond text‑only outputs.
Also developing:
Adobe Summit 2026 – AI Has Gone From Know-It-All to Actually Getting Work Done, Says Nvidia CEO
At Adobe Summit 2026 in Las Vegas, Adobe unveiled CX Enterprise and AI‑driven assistants that turn its creative suite into an agentic platform. Nvidia CEO Jensen Huang joined the keynote, declaring that AI has moved from merely “knowing it all” to actually completing work for users. He highlighted how AI agents already boost productivity in fields like radiology and software engineering, and previewed Nvidia’s Omniverse as the bridge to physical‑world actions. The announcements signal a coordinated push by Adobe and Nvidia to embed intelligent agents across SaaS and design tools.
Research Bits: Apr. 21
Researchers at the University of Michigan demonstrated a compute‑in‑memory (CIM) implementation of state‑space models using a 65 nm CMOS resistive‑RAM crossbar, achieving vector‑matrix multiplication within 4.6 bits of the ideal result while dramatically cutting energy use. In Tokyo, scientists from the Institute...
Huawei AI Glasses Take on Meta Ray-Ban and Oakley
Huawei unveiled its AI Glasses in China on April 20, pricing the base models at ¥2,499 (about $365) and a sunglasses variant at ¥2,899 ($420). The eyewear runs HarmonyOS, features a proprietary AI‑powered chip, a 1/2.8‑inch camera, and a built‑in real‑time...
You Don’t Have an AI Problem, You Have a Skills Problem
Artificial intelligence has moved from experiment to everyday tool, with 61% of UK organisations now allowing employees to use generative AI at work. Yet productivity gains are falling short—research shows firms are missing roughly 40% of the potential upside because...
The Myriad Complex Ways Young People Use AI
The Rithm Project surveyed 2,400 U.S. youths aged 13‑24, uncovering four broad AI‑usage groups and nine detailed portraits. Roughly 28% use AI infrequently, 39% for information and tasks, 18% for personal support, and 15% engage with humanlike AI characters. The...
Google Brings Pomelli in English to Small Businesses in Europe.
Google has launched Pomelli, its AI‑powered marketing tool, in English across the European Union, Iceland, Liechtenstein, Norway, Switzerland and the United Kingdom. Developed by Google Labs in partnership with DeepMind, Pomelli scans a small business’s website, proposes campaign ideas and...
Introducing AI Fraud Detection Agent
Gem launched its AI Fraud Detection Agent, the third AI tool in its recruiting suite, to automatically assess inbound applications for fraud risk. The agent evaluates six data signals—resume metadata, email, phone, LinkedIn, person verification, and IP/device—and assigns a high,...
10 Al Prompts For Building Viral Al Video Channel with with Gemini
The post outlines ten ready‑to‑use Gemini prompts that take a creator from concept to a viral AI‑driven short‑form video channel in 30 days. Each prompt tackles a specific stage—niche scouting, visual identity, hook creation, scripting, workflow, algorithm decoding, calendar planning,...
Intelligence, Rearranged: How Agents Are Changing Legal Work
Harvey’s new long‑horizon AI agents move beyond checklist‑driven tasks to self‑directed reasoning, allowing them to plan, research, iterate, and deliver complex legal work autonomously. The platform now runs over 700,000 agentic tasks daily and extracts more than 50 million terms each...
Google Launches Personal Intelligence in the Gemini App in the Arab World
Google has extended its Personal Intelligence feature to the Arab world through the Gemini app. The AI‑driven tool links data from Gmail, Photos and other apps, allowing Gemini to answer personalized queries such as travel itineraries or book recommendations. Users...
AI Automates Job Applications, Turning Effort Into Strategy
Stop sending 200 identical CVs. This might be the turning point in how people apply for jobs. A new open-source tool powered by Claude Opus 4.7 can now handle your entire job search workflow: • Finds relevant job openings automatically • Rewrites your CV...
Responsible AI in Practice: 2025 Global Insights From the AI Company Data Initiative
UNESCO’s 2025 Responsible AI Practice report highlights a widening gap between the rapid integration of artificial intelligence in corporate products and services and the slower evolution of governance and disclosure standards. Drawing on the Thomson Reuters Foundation’s AI Company Data...
Positive Reinforcement Speeds up AI Learning
You have seen sycophantic AIs. But with the new AI's coming if YOU kiss the ass of your agents they learn faster. I find I'm learning to be nicer to my digital friends. Reward them. Say "great job." That actually helps them...
Australian Operators Expand AI Deployment Across Operations - ACMA
Australian telecom operators are rapidly embedding artificial intelligence across core functions, according to a new Australian Communications and Media Authority (ACMA) report. The study notes that Telstra, Optus, TPG Telecom and Superloop have moved beyond isolated pilots to integrate AI...
LexisNexis and Luminance Announce Partnership to Embed Protégé Inside Luminance’s Contract Platform
LexisNexis and Luminance announced a partnership that embeds the Protégé AI assistant inside Luminance’s contract‑negotiation platform. The integration lets in‑house legal teams ask legal questions directly through Luminance’s natural‑language interface, Lumi, and receive AI‑generated answers. A migration path to Lexis+...
Whose Water Is AI Drinking in India?
Tech giants and Indian conglomerates have pledged $167.5 bn to build AI data centres across India. These facilities, concentrated in Mumbai, Bengaluru, Chennai, Hyderabad and Delhi‑NCR, consume massive water volumes—about 150 bn litres in 2025 and projected to exceed 350 bn litres by...
AI Elevates Financial Analysts to Strategic Decision Makers
AI is transforming the Financial Analyst role by automating repetitive tasks and improving efficiency in data analysis. It enables analysts to generate faster insights, build smarter forecasts, and enhance decision-making with advanced tools. Rather than replacing professionals, AI is shifting...
AI and Globalization Threaten White‑Collar Jobs and Welfare State
Yes indeed, upcoming “revolt”: my 2019 OUP book calls it the “Globotics Upheaval” 👉”Computing power is dissolving humans' monopoly on thinking, enabling AI-trained computers to compete for many of the same white-collar jobs. 👉”The combination of globalization and robotics is creating the...
How Can We Make AI Accountability Something Systems Can Actually Do?
AffectLog was founded to bridge the gap between high‑level AI ethics frameworks and the day‑to‑day reality of EdTech deployments. The startup delivers a federated, privacy‑preserving risk‑analytics suite that computes compliance against more than 300 regulatory constraints, including the EU AI...
AI in Education #4: Five Things I Learned From Carl Hendrick About Why This Time It Really Is Different
In the fourth episode of the AI in Education podcast, Carl Hendrick argues that AI is finally poised to transform schooling because it is built on solid learning science rather than hype. He highlights how modern AI can collect granular...
Qanooni AI and Docusign to Connect Legal AI with Intelligent Agreement Management
Qanooni AI and DocuSign announced a technology partnership that embeds Qanooni’s legal‑AI capabilities into DocuSign’s Intelligent Agreement Management suite. The integration lets legal teams draft, review, research and execute contracts from Microsoft Word or Outlook while leveraging DocuSign’s eSignature, Navigator...
White House Unveils National AI Legislative Framework
On March 20, the White House released a national artificial‑intelligence legislative framework outlining seven objectives, from mandatory parental controls to preempting state AI laws. The proposal urges Congress to codify child‑safety safeguards, ratepayer protections, IP licensing options, and regulatory sandboxes while...
Project Agora Brings Realize to North Africa, Expands Access to AI-Driven Performance Advertising
Project Agora has launched the AI‑driven performance platform Realize across North Africa, covering Egypt, Morocco, Algeria and Tunisia. The service taps a network of more than 11,000 premium publishers and offers format‑agnostic ad placements, AI‑optimized bidding and predictive audience models....
Hmmm: Rail-Optimized Networking for AI Workloads
Phil Gervasi’s recent piece promotes a “rail‑optimized” networking approach for AI training workloads, describing a mapping of endpoints to a dedicated plane within a standard leaf‑spine fabric. The article argues that keeping traffic within leaf switches and using server‑bus paths...
Philippe Pouletty, Carvolix
Physician‑entrepreneur Philippe Pouletty, founder of Carvolix, is launching AI‑driven tools for cardiology, including a decision‑support system and a robot that places transcatheter heart valves. The company also plans a robotic device to remove brain clots. Pouletty previously founded Abivax, a...
Advanced AI Raises Security Risks
Palo Alto Networks warns that emerging generative AI models will soon become powerful tools for cyber attackers. In internal tests, the firm’s AI completed the equivalent of a year’s penetration testing in just three weeks and demonstrated the ability to...
Harnessing AI Through a Global Finance Hub
Indonesia is positioning itself as a global AI hub, aiming to embed artificial intelligence across its 65 million small and medium enterprises. The article argues that AI can offset the country’s historically low productivity and leverage its demographic advantage. To realize...
AI-Industry Partnerships Red
AI is reshaping higher education—not just in theory, but in practice. The collaboration between SJTU and Huawei shows what this looks like: ⚙️ AI-driven research at scale 🎓 hands-on, project-based learning 🏫 fully digitalized smart campuses At the core: the Zhiyuan-1 platform + an AI...
Who Is Manufacturing the Faces We Trust?
Indian advertisers and brands are rapidly adopting AI‑generated digital spokespersons that can speak regional languages, adjust facial features, and deliver consistent messages without fatigue. Mankind Pharma announced a partnership to embed these generative systems across its marketing and multilingual communications,...
Turn Meeting Notes Into Polished Docs with AI
How my team used AI today: 1. We review doc in meeting and I give feedback. 2. I offer to edit/rewrite. 3. I feed Claude the meeting notes and ask to rewrite doc based on my feedback. 4. I review then share it out. 5. Team member likes it and...
Rise of Physical AI – KUKA and NVIDIA Partner to Launch Automation 2.0
KUKA and NVIDIA announced a strategic partnership at NVIDIA GTC, unveiling the KUKA Automation Management Platform (KUKA AMP) that blends traditional rule‑based control with intent‑based AI. The collaboration marks the launch of what KUKA calls Automation 2.0, a Physical AI framework that...
Zapier MCP Vs. Zapier SDK: What's the Difference?
Zapier introduced two distinct products for AI‑driven automation: Zapier MCP, a no‑code connector that lets chat and coding agents invoke pre‑built actions, and Zapier SDK, a TypeScript library that gives developers programmatic, type‑safe access to the same ecosystem. MCP works...
Zapier SDK: Run App Actions Directly From Your Code
Zapier has launched an open‑beta SDK that lets AI coding agents invoke over 9,000 pre‑built app integrations and an additional 3,000 via Relay. The toolkit bundles authentication, token refresh, retries and TypeScript‑generated types, removing the need for developers to hand‑code...
Seven Mental Models to Understand the AI Compute Era
The AI compute landscape expanded dramatically, with total tracked capacity jumping 8.5‑fold between Q1 2024 and Q4 2025, rising from 2.5 million to 21.3 million H100‑equivalent units. The article argues that this surge reflects a deeper power‑infrastructure race rather than merely more chips or...
The Revenue Impact of Human-Centered AI
Sales teams are inundated with AI dashboards that sit outside the flow of work, limiting impact. Highspot’s human‑centered AI, embodied in Deal Agent and AI Role Play, embeds real‑time insights and guided practice directly into sellers’ workflows. While 77% of...
Why AI Safety Should Be For-Profit?
The piece argues that AI safety should move from nonprofit‑driven research to for‑profit enterprises, using recent scandals—xAI’s Grok deepfakes, Character.AI’s teen‑suicide lawsuits, and OpenAI’s wrongful‑death claims—as proof that safety only improves under financial or legal pressure. It likens the emerging...
AI Companions Developed for Lonely Students in Australia
Researchers at the University of New South Wales have built two prototype AI companions, Tom and Mia, to help university students cope with loneliness. The chatbots, designed in English and Mandarin, were co‑created with Chinese student input and incorporate safeguards...
27 Physical AI Startups Secure $50M+ Funding, Hiring
list of 27 physical AI startups which raised > $50M in Q1 2026 , all likely hiring. Credit: Ivan Landabaso's "Startup Riders" website.
Glossy Research: 3 Out of 4 Brand Leaders Are Using AI for Data Analysis, but ROI on AI Spend Remains...
Glossy Research reports that three‑quarters of top fashion and beauty brands now rely on AI for data analysis, yet the financial returns on those investments remain modest. The study highlights an average ROI of under 10% per year, indicating early‑stage...
Will the UK’s Datacentre Strategy Deliver?
The UK’s Department for Science, Innovation and Technology (DSIT) has set a 2030 goal of delivering 6 GW of AI‑capable datacentre capacity. Recent analysis shows current build‑out is lagging, making the target unlikely without rapid acceleration in power and cooling investments....
How GAO’s Artificial Intelligence Use Cases Show What Practical Government AI Looks Like
The U.S. Government Accountability Office (GAO) released a March 2026 catalog listing ten AI use cases, five of which are fully operational. The tools automate search, text classification, document conversion, and legislative mandate analysis, while a generative AI prototype assists staff...
How ThredUp Is Using AI to Improve Its Marketing Operations and Move Further up the Funnel
ThredUp, the online resale platform, is embedding artificial intelligence across its operations, from automated product tagging to personalized recommendation engines. In its marketing division, AI is now being used to sharpen demand forecasting, allowing the team to allocate spend more...
Opus 4.6 Erroneously Writes Dummy Output to /Dev/Null
As I’m using @AnthropicAI Opus 4.6 on @aws Kiro CLI right now it keeps writing dummy to screen and tries to write to /dev/null all the time now. It suggests that dummy is a tool name and cannot tell me...
Can Perfectionist John Ternus Steer Apple Through the AI Age?
Apple’s long‑time hardware engineer John Ternus is set to replace Tim Cook as CEO on September 1, bringing a product‑first philosophy to a company that has lagged in AI. Ternus stresses leveraging technology to ship great products rather than shipping technology...
AI Won’t Cause Job Apocalypse, But Triggers Massive Role Shifts
There will be no “jobs apocalypse” due to AI — but there will be job chaos. @Gartner_inc 2025 AI Job Impacts Analysis found that starting in 2028-2029, AI will create more jobs than it eliminates. Yet, each year, over 32...
Why Equity Researchers Should Switch to Claude Now
I'm late to the party, but there are good reasons to switch to Claude, if you're in equity research: https://t.co/ZBPWGCeQZl
Half of Audience Will Wear Smart Glasses by Next Year
#Money2020 ANT Group Garry Sien great talk on AI he says half of this audience will be wearing smart glasses this time next year. I certainly will be. https://t.co/3fCBQFFCp0
IT’s Future: Humans and AI Working Together
The future of IT is a partnership between humans and AI. #AI #Innovation #Workforce #BusinessStrategy https://t.co/HxOiCAH48y
AI Design Tool Bridges Codex's Frontend Skill Gap
I feel like Codex's gap in frontend design skills can be easily made up if you use an AI design tool. My favorite is @tomkrcha's Pencil
Small Threads Spark Massive AI Agent Revolutions
This is an iconic thread that started a massive revolution and a complete rethinking around agents and smart harnesses around models and I mean a revolution. And as of right now it only has 25k views. Revolutions always start small.