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:
Instawork Launches Robotics Lab, Shifts Gig Platform to AI Data Provider
Instawork unveiled the Instawork Robotics Lab and a certification program that has already credentialed more than 20,000 gig workers, positioning the company as a major supplier of real‑world training data for robots. The move tackles the industry’s “100,000‑year” data gap and opens a new revenue stream for the San Francisco‑based marketplace.
China’s Robot Surge Highlights U.S. AI and Data‑Analytics Gap
China’s aggressive rollout of factory‑floor and humanoid robots is outpacing the United States, exposing a shortfall in American data‑analytics and AI implementation. Experts say the gap stems from fragmented U.S. policy and a lack of coordinated national strategy, while Beijing’s...
Cal.com Closes Core Codebase Over AI Security Fears, Launches MIT‑Licensed Cal.diy
Cal.com announced it is moving its production codebase behind closed doors, warning that AI tools like Anthropic’s Claude Mythos make open code a security liability. The company, which has raised more than $30 million, simultaneously unveiled Cal.diy, an MIT‑licensed, feature‑lite variant...
ZoomInfo Partners with Pinecone to Power AI Contact Recommendations, Lifting Engagement 50%
ZoomInfo announced a partnership with Pinecone to embed the latter's serverless vector database into its sales intelligence platform. The integration powers real‑time AI contact recommendations, delivering a 50% rise in user engagement and a two‑fold boost in relevance, while handling...
Capacity Efficiency at Meta: How Unified AI Agents Optimize Performance at Hyperscale
Meta’s Capacity Efficiency Program has deployed a unified AI agent platform that automates both detection and remediation of performance regressions and optimization opportunities across its hyperscale infrastructure. By encoding senior engineers’ domain expertise into reusable skills and standardized tool interfaces,...
Maine Enacts First AI Disclosure Law for Political Ads
Maine lawmakers approved LD 517, mandating that political campaigns and PACs label ads generated with artificial intelligence. The law, signed by Gov. Janet Mills, marks the nation’s first AI‑disclosure requirement for political advertising, aiming to curb misinformation and restore voter...
Launch AI Pitch Coach Now, Refine Later
I built chatbot and voice versions of myself so founders can get feedback without waiting for my calendar. It stress-tests decks and filters the strongest pitches to my team. Was just featured byPitchBook. The tech has gaps but waiting for perfection...
Google Adds Nano Banana-Powered Image Generation to Gemini’s Personal Intelligence
Google announced that its Gemini Personal Intelligence feature will now generate images using the Nano Banana engine, which draws on a user’s Google account data such as Gmail and Photos. The AI can infer preferences and relationships without explicit prompts,...
Roblox’s AI Assistant Gets New Agentic Tools to Plan, Build, and Test Games
Roblox announced a suite of agentic features that upgrade its Roblox Assistant from a simple prompt‑response tool to a collaborative development partner. The new Planning Mode lets the assistant analyze code, ask clarifying questions, and generate editable action plans before...
Nano Banana Can Now Make Personalized AI Images Based On Your Photos Library
Google has rolled out Gemini Personal Intelligence in Nano Banana 2, letting the AI image generator tap users' Google Photos libraries to create personalized pictures without manual prompts or uploads. The feature leverages photo labels and metadata, while Google assures the private...
New Ways to Create Personalized Images in the Gemini App
Google’s Gemini app now leverages Personal Intelligence and Nano Banana 2 to generate custom images directly from a user’s Google Photos library and inferred interests. Subscribers can issue simple, natural‑language prompts and the system automatically pulls relevant personal context, eliminating the need...
Teens Alarmed at What AI Is Doing to Their Minds
A Drexel University study examined 318 Reddit posts from teens using the AI chatbot Character.AI and found clear signs of behavioral addiction. The research identified all six classic addiction factors—salience, withdrawal, tolerance, relapse, mood modification, and conflicting desires—despite users’ growing...
Super Hi-Fi Introduces AI Music Scheduling
Super Hi‑Fi unveiled Neuron, an AI‑powered music‑scheduling engine that leverages neuroscience research to reshape how songs are sequenced for listeners. The platform replaces decades‑old static rule‑sets with dynamic optimization of attention, contrast, novelty and reward, aiming to boost Time Spent...
Planview and Ozmosi: Bringing AI-Powered Insight to Pharmaceutical Portfolio Decisions
Planview has partnered with Ozmosi to embed curated clinical‑trial and market intelligence into its portfolio management platform, enhancing AI‑driven decision‑making for pharmaceutical companies. The integration supplies real‑time external signals, allowing teams to benchmark assumptions, identify pipeline gaps, and monitor competitive...
Check Fraud in 2026: Why AI Is Now Essential
Check fraud remains a major threat, representing about 30% of all fraud losses in the United States, second only to debit‑card fraud. Reports of suspicious activity tied to check fraud surged 90% from 2021 to 2023, and high‑profile cases involving...
Peak Technologies Partners with Jacobi Robotics to Deliver Next-Generation Mixed-Case Palletizing Automation
Peak Technologies has teamed up with Jacobi Robotics to bring the OmniPalletizer, an AI‑driven mixed‑case palletizing system, to complex warehouses. The platform removes the need for upstream buffering, sorting and manual programming by using real‑time motion planning, computer vision and...
Never Used AI Seriously? Start Here.
The post outlines eight no‑code AI tools—Claude, Perplexity, Nano Banana Pro, Seedance 2.0, OpenClaw, NotebookLM, Claude (again for coding), and Gamma—each paired with a quick‑start guide and a pro tip. It frames these services as ready‑made productivity boosters for anyone without...
Hyper-Personal Shopping: The Rise of AI Shopper Bots
Retailers like John Lewis and Frasers are deploying AI‑powered shopper bots to deliver hyper‑personalized shopping experiences. These bots analyze real‑time browsing data and engage customers via chat or voice, offering curated product selections instantly. Early pilots have shown higher conversion...
Generative AI Data Scientists Replace Traditional Roles
RIP Data Scientists. The Generative AI Data Scientist is NOW what companies want. This is actually good news. Let me explain:
First Mass-Production Deployment of WeRide’s WRD 3.0 ADAS Solution in Co-Developed GAX Aion N60
GAC Aion has opened pre‑sales for the Aion N60, a smart urban SUV equipped with WeRide’s WRD 3.0 ADAS platform, formerly known as WePilot. The vehicle marks the first mass‑produced passenger car jointly developed by WeRide and GAC Aion and the...
Anthropic Launches Opus 4.7, Its Most Capable Model
Opus 4.7, Anthropic 's most capable model to date, is now live in bolt.new !
Teaching in the Age of AI: Join EdTechTeacher’s Free Expert Webinar Series This May
EdTechTeacher is hosting a free, three‑part webinar series in May to help K‑12 educators navigate AI in the classroom. The sessions, held on Wednesdays at 7 PM ET, focus on digital wellness, AI‑assisted writing, and integrating AI into math instruction. Each webinar...
Claude Opus 4.7 Executes Prompts Exactly, No Guesswork
claude opus 4.7 is here. obviously it's smarter. but there's some interesting new quirks and features with this model here's everything you should know: > opus 4.7 follows instructions extremely literally now. if your prompts were vague or loosely written, older models...
OpenAI Updates Agents SDK, Aims at Building Secure Agents
OpenAI released an updated Agents SDK featuring a model‑native harness that enables agents to operate across files, tools, and a new sandboxed execution environment. The SDK adds configurable memory and sandbox‑aware orchestration, simplifying secure agent deployment for enterprise developers. Analysts...
AI’s Energy Appetite Is Making Power Bills Harder to Swallow for Americans
A new American Home Shield survey of 1,003 consumers shows that 88% have seen utility costs rise, with the average electric bill now $187. Respondents living near data centers—14% of the sample—report a 94% incidence of higher bills and pay...
AI Makes SEO a Top Budget Priority—Will You Get Hired?
AI is the greatest thing that ever happened to SEO. Hear me out. It's suddenly the highest marketing priority. Demand is off the charts. You don't need to convince companies they should be doing this. They have genuine burning desire...
AI Succeeds when Decisions Have Clear Ownership
Most AI strategies fail for one reason: No one owns the decisions. Enterprises are scaling models. But not scaling decisions. What breaks: → Fragmented ownership → No execution governance → No orchestration What works instead: Decision systems. Enterprise AI leaders are shifting to: → Decision ownership layers ...
5 Considerations for Building an AI-Ready Infrastructure
The article outlines five essential considerations for higher‑education institutions building AI‑ready infrastructure, spanning consumption models, orchestration, facilities, governance, and scalable growth. It advises matching AI workloads to data sensitivity, ensuring container and GPU orchestration maturity, and verifying power, cooling, and...
AI Could Automate $13T Mortgage Market, Ditch Advisors
Wealthfront runs 45% EBITDA margins without a single advisor. Now David Fortunato wants to automate the $13 trillion mortgage market. (0:00) Can AI Replace Your Financial Advisor? (1:47) How Close Is Self-Driving Money? (3:59) What Happens to 400K Financial Advisors? (4:43) Is the 1%...
Testing Perplexity's Speed: Can It Overtake OpenClaw?
I’m going to be testing this. Perplexity is shipping at impressive speeds. But can they dethrone OpenClaw?
Nigeria’s Enugu State Plans AI Insitute in Bold Bet on Digital Talent Exports
Enugu State in southeastern Nigeria is launching a specialised artificial‑intelligence institute as the flagship of its "talent city" strategy, linking AI, cloud, cybersecurity and software engineering training directly to global outsourcing contracts. The project, budgeted at roughly $15 million, will repurpose...
Why AI Feels Different for Everyone, Even Same Tasks
Episode 3 of Wading Through AI is up! In this episode, @DemetriSpanos and I look at the aspects of AI technology that cause people to have wildly different experiences when using it, even for similar tasks: https://t.co/3GKLqTXCXL
Claude Opus 4.7 Feels Like a Promising Upgrade
Spending some time with my new friend Claude Opus 4.7. Already feels like an upgrade. Hope it doesn't go downhill like 4.6! https://t.co/ym6LY6lpWZ
Mythos Poses Risk to SEC Market-Tracking Database, Group Says
Anthropic’s new AI model Mythos could exploit the SEC’s Consolidated Audit Trail (CAT), a database that tracks every trade in U.S. equities. The American Securities Association warned that the model enables mass identity theft, portfolio exposure, and insider‑threat amplification, and...
Design AI in Education, Not Just Devotion
🧠Artificial Intelligence in #Education Needs Design, Not Devotion 🔥Join me next week at @sovaisv for a critical discussion on AI in education and how success depends on how learning is organized around it. https://t.co/jhNrMW0EUj @stacyherbert @JethroToro #AI @nayibbukele
AI's Relentless Pace Fuels a Volatile, Exciting Year
The rate of progress in AI is relentless. You can capture the upside of its volatility with @aisdk and @vercel AI Gateway. Congrats to @anthropicai on another banger ship, but @xai, @openai, and @googleai are coming. Gonna be a fun year.
Carrot Launches ‘Carrot Intelligence’ AI Platform for Global Fertility and Family Care
Carrot, a global fertility and family‑care platform, unveiled Carrot Intelligence, an AI engine built on a proprietary clinical dataset exceeding $1 billion in claims across 195 countries. The platform fuels a new Global Price Monitoring System that automatically spots billing anomalies,...
AI Becomes Essential; Non‑Adopters Lose Doctors
Very soon, in radiology and medical centers using AI, you will hear: “Do you remember when we didn’t have it ? How did we manage ?” Institutions without AI won’t just lag, they’ll struggle to attract and retain physicians.
App Now Fully Drivable by Agent – First Step
We decided to make the entire app drivable by agent This is the first step
Less Liability Could Solve the AI Chatbot Suicide Problem
AI chatbots are increasingly used for mental‑health support, with more than a million Americans turning to them weekly for anxiety, depression and relationship advice. Studies of Replika and other bots show users describe them as confidants, and dozens credit them...
AI Agents Require Complex, Multi‑Layered Development Effort
Creating and deploying AI agents aren’t exactly overnight tasks. AI agents still rely on a genAI foundation, as well as machine learning, statistical analysis, robotic process automation, intelligent document processing, and more. @forbes https://t.co/fKgjdw7W9W
Sweden's AI Ban Risks Missing Educational Design Benefits
🚨Is Sweden aversion to AI in education a tragic mistake? 👉Artificial intelligence in education needs design, not devotion. It's value in education depends on how learning is organized around it. 🔗New in @PsychToday https://t.co/jhNrMW0EUj #education #AI #learning
Optical Fiber Networks Can Keep Rail Networks Safe
Chinese researchers demonstrated that distributed acoustic sensing (DAS) on existing underground fiber‑optic cables can continuously monitor railway safety. AI models achieved 98.75% accuracy detecting faulty train wheels, 99.6% for broken sound barriers, and 97% for intrusions or debris. The approach...
Students Favor AI Over Basics, Resistance Rises
Students are using AI tools instead of building foundational skills - but resistance is growing https://t.co/TWqXiap2Ky via @ZDNET
Google, Pentagon Negotiate Secret AI Contract, Mending Military Ties
Google and Pentagon Discuss Classified AI Deal as Company Rebuilds Military Ties — The Information https://t.co/hHvbCxnlot
AI Models Keep Consistent Personalities Across Generations
Claude remains irreducibly Claude. If you know, you know. (The fact that models have distinct personalities that are consistent across generations is technically interesting, it also makes it very easy to use new releases when they come along, because they feel...
Empowerment in RL Linked to Bayesian Causal Models
New preprint of a paper with Eunice Yiu to appear in Philosophical Transactions A, Special issue: World models, 2026. The theoretical link between empowerment in RL and Bayesian causal models with cool new data. https://t.co/3sGo14sh1f
CEOs Must Adapt Strategically to AI Disruption
CEO themes and insights how does a CEO respond to AI business disruption? by @Timothy_Hughes https://t.co/LanW9VTJdY @DLAIgnite #SocialSelling #DigitalSelling #Sales #SalesTips #SalesLeader #Marketing #Leadership #ArtificialIntelligence #MarketingStrategy https://t.co/4SxQ0eSsnc
AI Now Outperforms Experts; We Must Decide Next Steps
2024: AI beat the human expert 1 in 10 times on real knowledge work. 2026: Opus 4.7 beats the human expert 3 out of 4 times. The question isn't whether AI is good enough. It's what we do about it. https://t.co/NN4skKcIuR
AllBirds Unexpectedly Joins
Admittedly, I did not have AllBirds on the AI LUMAscape on my bingo card... https://t.co/FJbnvzuvT1