Today's AI Pulse
Anthropic fixes three missteps that briefly dulled Claude Code
Anthropic disclosed three separate changes that temporarily reduced Claude Code's performance and restored the model by April 20. A March 4 switch lowered the default reasoning effort from high to medium, which was reverted on April 7, and a March 26 cache‑clearing bug erased session context, causing forgetfulness and higher token usage until fixed. The issues were fully resolved by April 20.
Also developing:
By the numbers: Tesla to acquire AI‑hardware firm for up to $2B
I Spent a Month Living with a $430 AI Pet, the Casio Moflin
Casio’s Moflin is a $430 AI‑enabled plush pet that learns simple emotional responses over weeks via the MofLife app, offers a “deep sleep” off mode, and converts audio into non‑identifiable data to distinguish voices rather than record conversations. A monthlong test finds Moflin convincingly lifelike in movement and social appeal—useful for children or memory care—and far cheaper than Sony’s $3,200 AIBO, but its AI feels closer to an advanced Furby than a fully interactive companion. Privacy concerns appear limited by design and network checks, yet the product’s high price and limited emotional sophistication raise questions about its value for mainstream buyers even as it may find a receptive market in Japan.
Sticking with Comet over OpenAI’s ChatGPT Atlas
I like exploring AI tools and new apps. So of course jumped at Comet. It is my primary browser and I love it. Now OpenAI has launched ChatGPT Atlas, which seems to be their version of Comet. So far I see...
Agentic Platform Race Heats Up with OpenAI Browser
I'm off to @rwang0's enterprise conference today. On the way out @flowith_ai just dropped me a note and a new thing to use that takes their agentic platform a lot further. More coming on Thursday from them. Can't talk about...
Google's New Vibe Coding AI Studio Experience Lets Anyone Build, Deploy Apps Live in Minutes
Google updated its AI Studio with a redesigned Build tab that lets novices and developers create, edit and deploy web apps in minutes using Gemini models and mix-and-match capabilities like Nano Banana, Veo, Imagine and Flashlight. The free-to-start “vibe coding”...
AI Development Demands Hypothesis‑First, Fast Iterative Launches
Every AI feature is a hypothesis, not a promise. That’s the biggest shift from SaaS to AI. In SaaS, you built toward predictability. You learned first, then launched because outcomes were deterministic. If you designed it right, it worked the same...
Retail’s Holiday 2025: More Sales, Profits, And AI-Driven Commerce
Retailers should expect a stronger Q4 2025 holiday season with year‑over‑year sales and profits rising as consumers continue to spend despite price worries and retailers operate leaner or use AI to cut costs. Promotions will remain muted through early November...
AI Video Generator Kling Shines at Cannes MIPCOM
Have you ever been to Cannes? I have. The audience there is the most powerful in the content business, so this caught my eye. @Kling_ai was there for the first time this year, significant because AI video generators haven't gotten that...
Safety and Innovation Thrive Together with Anthropic’s Constitutional AI
LIVE FROM DREAMFORCE: Paul Smith (@realpaulsmith), Chief Commercial Officer at @Anthropic, on balancing innovation and safety in AI. “We don’t want to be a bottleneck on innovation. You can have safe, trusted AI and still have incredible innovation, those two things...
YouTube’s Likeness Detection Technology Has Officially Launched
YouTube has officially launched its likeness detection tool to eligible creators in the YouTube Partner Program after a pilot, enabling creators to detect and request removal of AI‑generated videos that use their face or voice. Onboarding requires consent, ID and...
That All‑Too‑Familiar Moment We All Recognize
You know the one. That little bubble in the corner. "Hi! 👋 I'm here to help!" it chirps, right when you're trying to update a deal, export a report, or just... do your job. And what does it want? It...
WhatsApp and Messenger Add New Warnings to Help Older People Avoid Online Scams
Meta rolled out new scam‑detection warnings in WhatsApp and Messenger aimed at protecting older users, including alerts when unknown contacts request screen sharing and AI‑backed flags for suspicious Messenger messages. The company says it disrupted about 8 million scam accounts...
AI's Hidden Subconscious Mirrors Human Biases and Creativity
I didn't understand my own subconscious mind until I visited the VR/AR lab at Stanford University and Jeremy Bailenson had me walk the plank. There I learned that my subconscious mind would overrule my rational mind. In our newsletter this...
Stop Optimizing, Start Reinventing: Three Imperatives For Marketing Leaders in 2026
At MAICON 2025, marketing leaders were urged to move beyond incremental efficiency gains and pursue 10x reinvention by treating content as a structured, machine-readable data layer and designing for both human and AI agent audiences. Key imperatives include measuring system-level...
When Ideas Become Rules, Questioning Becomes Hard
Every product starts with an idea that works so well it turns into a rule. The longer that rule holds, the harder it becomes to question. For Crazy Egg, the rule was simple: keep analytics honest and easy to grasp....
Data Sovereignty in a VUCA World: Why Flexibility Matters More than Absolutes
Amid accelerating geopolitical, regulatory and technological change, SnapLogic AI and data expert Dominic Wellington argues that enterprises should prioritize flexibility over rigid data‑sovereignty absolutes to maintain resilience. He recommends a spectrum of delivery options (public/private cloud and on‑prem), vendor independence,...
Cultivate Human Skills to Thrive Amid AI Agents
AI Agents Are Coming For Your Job Tasks — Here’s How To Stay Ahead Automation is intensifying. This article shows how to position yourself for the future by focusing on uniquely human skills and aligning with how AI agents will...
Claude Code: Local AI Power Beyond Web Limits
More non-engineers should be learning to use Claude Code. Two things stopping them: 1) it’s in a terminal and terminals are icky for business folks (if you say “omg no it’s so easy”, it’s probably because you’re a dev) 2) Most use...
Wyze’s New Solar-Powered Pan-and-Tilt Security Camera Is Less than $80
Wyze launched the Solar Cam Pan, its first battery-powered, solar-charged 2K pan-and-tilt security camera, priced at $79.98—undercutting rivals like Reolink, Tapo and Eufy. The IP65-rated device offers 360° pan, 70° tilt, AI person detection with automatic tracking, IR night recording,...
Weekly Free AI Deep Dives, No Fluff
Every week, I share something exclusive (and completely free) that I don’t post anywhere else. New tools, research, lessons, and ideas shaping the future of AI - broken down simply, without the noise. If you like learning how things actually work, this...
Yelp’s AI Can Now Take Reservations over the Phone
Yelp launched two AI voice agents—Yelp Host for restaurants and Yelp Receptionist for broader businesses—that can answer calls, take and modify reservations, provide wait times, capture special requests, send menu links, and handle lead qualification and scheduling. Host is priced...
How to Brace Your Business Infrastructure for the Impact of ChatGPT-5
OpenAI’s ChatGPT‑5 advances to ‘PhD‑level’ reasoning with a unified model and a one‑million‑token context window, enabling enterprises to process larger datasets, automate multi‑step workflows and extract richer insights. Those gains come with heavy infrastructure demands—heightened bandwidth, compute and latency sensitivity—that...
Vision Action Models Accelerate Real‑Time AI Video and Robotics
AI video is doing same as Tesla Self Driving: upgrading quickly. The robotics entrepreneurs in San Francisco told me about VLAs today. Vision Action Models. The AI can now understand the world in real time. Important for robotics. And same kind of...
Building AI-Ready Cultures in Life Sciences R&D - with Xiong Liu of Novartis
In this episode of the AI in Business podcast, Xiong Liu, Director of Data Science and AI at Novartis, discusses how generative AI and foundation models are reshaping life‑sciences R&D, from clinical workflows to collaborative research. He emphasizes the need...
AI’s Financial Blind Spot: Why Long-Term Success Depends on Cost Transparency
Businesses are racing to deploy AI but lack of cost transparency risks turning promising projects into expensive failures, argues Apptio. With 68% of tech leaders planning bigger AI budgets and 39% expecting AI to drive future budget growth—despite an average...
OpenAI Announces ChatGPT Atlas, an AI-Enabled Web Browser to Challenge Google Chrome
OpenAI launched ChatGPT Atlas, an AI-enabled web browser now available on macOS with Windows, iOS and Android support coming soon, and CEO Sam Altman set to formally unveil it in a livestream. Atlas positions OpenAI to challenge Google Chrome and...
The Unexpected Benefits of AI PCs: Why Creativity Could Be the New Productivity
New research and industry data position AI-enabled PCs—laptops with on-device neural processing—as catalysts for creativity-driven productivity, with MIT Sloan and HP-backed findings showing generative AI can enhance human creativity when workers have the right tools. Early adopters report measurable gains:...
Optimize Your Content for LLMs and AI
What is #LLMO? Optimize content for AI & large language models! https://t.co/6Jdoo8FXds #LLMs #LLM #genai #GenerativeAI #AI #artificial_intelligence #SEO #searchengine @SpirosMargaris @PawlowskiMario @mvollmer1 @gvalan @ipfconline1 @LaurentAlaus @Shi4Tech @Fisher85M @kalydeoo @Ym78200 @Nicochan33 @chboursin @3itcom @Fabriziobustama @sallyeaves @helene_wpli @ahier @rwang0 @EvanKirstel @RLDI_Lamy @Analytics_699 @Khulood_Almani...
Turn Papers Into Production-Ready Code in Minutes
Go from research paper → codebase in minutes. DeepCode turns prompts, PDFs, or URLs into complete production-grade codebases: with backend, frontend, even algorithmic logic! 🔥 100% open-source. Repo in 🧵↓ https://t.co/jNbLhRqlH2
Why an AGI Delay Doesn't Mean an AI Bubble
Nathaniel Whittemore argues that a slowdown in achieving artificial general intelligence does not signal an AI bubble, pointing to concrete industry shifts that show AI’s transformative impact. He draws on consulting trends—lower delivery costs, evolving client expectations, and the rise...
ChatGPT Atlas: AI‑First Browser Launches on macOS
Our new AI-first web browser, ChatGPT Atlas, is here for macOS. Please send feedback! Availability on other platforms to follow.
OpenAI Launches AI‑first Web Browser for Innovators
OpenAI introducing a new Web browser for AI first thinkers.
Bryan Cranston and SAG-AFTRA Say OpenAI Is Taking Their Deepfake Concerns Seriously
Bryan Cranston, OpenAI, SAG-AFTRA and major talent agencies issued a joint statement saying OpenAI has “strengthened guardrails” around Sora 2 after unauthorized deepfakes of Cranston—including a viral image of him taking a selfie with Michael Jackson—appeared on the app. OpenAI...
ChatGPT Atlas Launches as macOS Browser
ChatGPT Atlas. Macos browser https://t.co/1jdCCZN5Gl
Deepseek's OCR Model Hits Trend, 100K HuggingFace Followers
Deepseek released a new top trending open-source OCR model and paper and crossed 100,000 AI builder followers on @huggingface (from less than 5,000 beginning of the year)! https://t.co/wpYDD3HnE9
Unlock the Power of Images with AI Sheets
Hugging Face updated its open‑source AI Sheets tool to add full vision support, letting users view, extract, analyze, generate and edit images directly inside spreadsheets using thousands of open models via Inference Providers. The release enables tasks from receipt line‑item...
Only Path to Shippable Code From AI Agents?
Is this the only way to get coding agents to produce shippable quality code? https://t.co/bZvxMN6JEv
Elsevier Pioneers Decade-Long Responsible AI for Research
Elsevier has been using artificial intelligence responsibly for more than a decade to develop innovative solutions that support researchers, research decision-makers and their institutions. https://t.co/eZ56HGkCmR #AI #Research #Elsevier #ResponsibleAI
Supercharge Your OCR Pipelines with Open Models
A new practical guide maps the rapidly evolving landscape of open‑weight vision‑language OCR models, explaining when to fine‑tune versus use off‑the‑shelf models and how to move beyond basic transcription to multimodal retrieval and document QA. It compares leading open models...
Tesla's Self-Driving Beats Waymo, Headlights Upgrade
Holy shit. @tesla Self Driving. Wow. @WholeMarsBlog was right. Even the matrix headlights improved. Waymo is way inferior. https://t.co/FpdBKQlpMB
DeepSeek OCR: Beyond Text Extraction with AI
DeepSeek OCR, More Than OCR! https://t.co/pIoCgsugxO #OCR #deepseek #deepseekocr #LLMs #genai #AI #GenerativeAI #artificial_intelligence @AlbertoEMachado @Eli_Krumova @postoff25 @Khulood_Almani @anand_narang @NutritiousMind @baski_LA @TanyaSinha_ @devaang @AlAmadi1 @jeancayeux @enilev @efipm @mvollmer1 @Nicochan33 @RagusoSergio @FrRonconi @Shi4Tech @sallyeaves @LaurentAlaus @Fabriziobustama @smaksked @MargaretSiegien @PawlowskiMario @gvalan @Ym78200 @mikeflache...
AI Infrastructure with Flex AI CEO Brijesh Tripathi
The post surveys recent AI‑infrastructure news and interviews, highlighting Amazon’s lawsuit against Perplexity, OpenAI’s $38 billion AWS partnership, Nvidia’s $5 trillion valuation, and emerging AI applications from enterprise workflow to defense and digital legacy. It underscores how shifting cloud alliances, regulatory scrutiny,...
Nicholas Thompson, CEO of The Atlantic, on Leading The Atlantic Through the “AI Hurricane Coming Through”
In a recent episode of the *In AI We Trust?* podcast, EqualAI CEO Miriam Vogel interviews Nicholas Thompson, CEO of *The Atlantic*, about navigating the looming "AI hurricane" and the evolving role of journalists in AI literacy. Thompson outlines his...
GPT‑6 Unlikely This Year; GPT‑5 Will Redefine Interaction
GPT-6 by the end of the year? "And honestly, I don't think it's going to be happening. It would be very weird to have this massive launch GPT5 really a fundamental shift in the way users interact with chat GPT...
New Paper Defines AGI by Human‑Level Cognitive Versatility
What even is AGI anymore? A new paper attempts to define it. "So our definition of AGI is an AI that can match or exceed the cognitive versatility and proficiency of a well-educated adult. To measure this, we assess multiple...
AI Proficiency Requires Trainable Mindset, Not Just Technical Skill
Most leaders assume technical competence = AI proficiency. Wrong - it's a separate, trainable skill that needs explicit development across the org. And some people are primed for that mindset. Our AI courses have trained hundreds of thousands of business professionals...
Text Diffusion: Simple Bi‑Directional Transformer Beats Autoregression
Nice, short post illustrating how simple text (discrete) diffusion can be. Diffusion (i.e. parallel, iterated denoising, top) is the pervasive generative paradigm in image/video, but autoregression (i.e. go left to right bottom) is the dominant paradigm in text. For audio I've...
AI Amplifies Bad Data—Clean Foundations First
Most teams dream of using AI to save them from complexity. The dashboards, the messy reports and the endless debates over numbers. Everyone wants a magic switch. The reality is less cinematic. AI doesn’t clean up the mess for you. It...
Veo 3.1 Dominates Leaderboards with Massive Performance Gains
Awesome to see Veo 3.1 top the LMArena video leaderboards by a large distance with big improvements over Veo 3.0 for text-to-video (+30) and image-to-video (+70)! 🔥Huge congrats to the team! Try it for yourself in https://t.co/QgTpxTL8DQ and the @GeminiApp
Fuel Your OODA Loop for AI-Driven Competition
Quick insights on OODAcon and how we will help add jet fuel to your OODA Loop, with @junaidislam. Bottom line: This is about ensuring you are ready to orient yourself to compete and win in the new AI enabled world....
AI Isn't a Bubble, Unlike the Dot‑Com Era
this time last year post-@stateofaireport we heard lots of “is ai a bubble” chatter, just like we do today we wrote a piece on @airstreetpress arguing that ai is not like the dot com thinking of writing up a v2...